Tag Archives: AI for Social Good

Simulations illuminate the path to post-event traffic flow

Fifteen minutes. That’s how long it took to empty the Colosseum, an engineering marvel that’s still standing as the largest amphitheater in the world. Two thousand years later, this design continues to work well to move enormous crowds out of sporting and entertainment venues.

But of course, exiting the arena is only the first step. Next, people must navigate the traffic that builds up in the surrounding streets. This is an age-old problem that remains unsolved to this day. In Rome, they addressed the issue by prohibiting private traffic on the street that passes directly by the Colosseum. This policy worked there, but what if you’re not in Rome? What if you’re at the Superbowl? Or at a Taylor Swift concert?

An approach to addressing this problem is to use simulation models, sometimes called "digital twins", which are virtual replicas of real-world transportation networks that attempt to capture every detail from the layout of streets and intersections to the flow of vehicles. These models allow traffic experts to mitigate congestion, reduce accidents, and improve the experience of drivers, riders, and walkers alike. Previously, our team used these models to quantify sustainability impact of routing, test evacuation plans and show simulated traffic in Maps Immersive View.

Calibrating high-resolution traffic simulations to match the specific dynamics of a particular setting is a longstanding challenge in the field. The availability of aggregate mobility data, detailed Google Maps road network data, advances in transportation science (such as understanding the relationship between segment demands and speeds for road segments with traffic signals), and calibration techniques which make use of speed data in physics-informed traffic models are paving the way for compute-efficient optimization at a global scale.

To test this technology in the real world, Google Research partnered with the Seattle Department of Transportation (SDOT) to develop simulation-based traffic guidance plans. Our goal is to help thousands of attendees of major sports and entertainment events leave the stadium area quickly and safely. The proposed plan reduced average trip travel times by 7 minutes for vehicles leaving the stadium region during large events. We deployed it in collaboration with SDOT using Dynamic Message Signs (DMS) and verified impact over multiple events between August and November, 2023.

One policy recommendation we made was to divert traffic from S Spokane St, a major thoroughfare that connects the area to highways I-5 and SR 99, and is often congested after events. Suggested changes improved the flow of traffic through highways and arterial streets near the stadium, and reduced the length of vehicle queues that formed behind traffic signals. (Note that vehicles are larger than reality in this clip for demonstration.)

Simulation model

For this project, we created a new simulation model of the area around Seattle’s stadiums. The intent for this model is to replay each traffic situation for a specified day as closely as possible. We use an open-source simulation software, Simulation of Urban MObility (SUMO). SUMO’s behavioral models help us describe traffic dynamics, for instance, how drivers make decisions, like car-following, lane-changing and speed limit compliance. We also use insights from Google Maps to define the network’s structure and various static segment attributes (e.g., number of lanes, speed limit, presence of traffic lights).

Overview of the Simulation framework.

Travel demand is an important simulator input. To compute it, we first decompose the road network of a given metropolitan area into zones, specifically level 13 S2 cells with 1.27 km2 area per cell. From there, we define the travel demand as the expected number of trips that travel from an origin zone to a destination zone in a given time period. The demand is represented as aggregated origin–destination (OD) matrices.

To get the initial expected number of trips between an origin zone and a destination zone, we use aggregated and anonymized mobility statistics. Then we solve the OD calibration problem by combining initial demand with observed traffic statistics, like segment speeds, travel times and vehicular counts, to reproduce event scenarios.

We model the traffic around multiple past events in Seattle’s T-Mobile Park and Lumen Field and evaluate the accuracy by computing aggregated and anonymized traffic statistics. Analyzing these event scenarios helps us understand the effect of different routing policies on congestion in the region.

Heatmaps demonstrate a substantial increase in numbers of trips in the region after a game as compared to the same time on a non-game day.
The graph shows observed segment speeds on the x-axis and simulated speeds on the y-axis for a modeled event. The concentration of data points along the red x=y line demonstrates the ability of the simulation to reproduce realistic traffic conditions.

Routing policies

SDOT and the Seattle Police Department’s (SPD) local knowledge helped us determine the most congested routes that needed improvement:

  • Traffic from T-Mobile Park stadium parking lot’s Edgar Martinez Dr. S exit to eastbound I-5 highway / westbound SR 99 highway
  • Traffic through Lumen Field stadium parking lot to northbound Cherry St. I-5 on-ramp
  • Traffic going southbound through Seattle’s SODO neighborhood to S Spokane St.

We developed routing policies and evaluated them using the simulation model. To disperse traffic faster, we tried policies that would route northbound/southbound traffic from the nearest ramps to further highway ramps, to shorten the wait times. We also experimented with opening HOV lanes to event traffic, recommending alternate routes (e.g., SR 99), or load sharing between different lanes to get to the nearest stadium ramps.

Evaluation results

We model multiple events with different traffic conditions, event times, and attendee counts. For each policy, the simulation reproduces post-game traffic and reports the travel time for vehicles, from departing the stadium to reaching their destination or leaving the Seattle SODO area. The time savings are computed as the difference of travel time before/after the policy, and are shown in the below table, per policy, for small and large events. We apply each policy to a percentage of traffic, and re-estimate the travel times. Results are shown if 10%, 30%, or 50% of vehicles are affected by a policy.

Based on these simulation results, the feasibility of implementation, and other considerations, SDOT has decided to implement the “Northbound Cherry St ramp” and “Southbound S Spokane St ramp” policies using DMS during large events. The signs suggest drivers take alternative routes to reach their destinations. The combination of these two policies leads to an average of 7 minutes of travel time savings per vehicle, based on rerouting 30% of traffic during large events.


This work demonstrates the power of simulations to model, identify, and quantify the effect of proposed traffic guidance policies. Simulations allow network planners to identify underused segments and evaluate the effects of different routing policies, leading to a better spatial distribution of traffic. The offline modeling and online testing show that our approach can reduce total travel time. Further improvements can be made by adding more traffic management strategies, such as optimizing traffic lights. Simulation models have been historically time consuming and hence affordable only for the largest cities and high stake projects. By investing in more scalable techniques, we hope to bring these models to more cities and use cases around the world.


In collaboration with Alex Shashko, Andrew Tomkins, Ashley Carrick, Carolina Osorio, Chao Zhang, Damien Pierce, Iveel Tsogsuren, Sheila de Guia, and Yi-fan Chen. Visual design by John Guilyard. We would like to thank our SDOT partners Carter Danne, Chun Kwan, Ethan Bancroft, Jason Cambridge, Laura Wojcicki, Michael Minor, Mohammed Said, Trevor Partap, and SPD partners Lt. Bryan Clenna and Sgt. Brian Kokesh.

Source: Google AI Blog

Open sourcing Project Guideline: A platform for computer vision accessibility technology

Two years ago we announced Project Guideline, a collaboration between Google Research and Guiding Eyes for the Blind that enabled people with visual impairments (e.g., blindness and low-vision) to walk, jog, and run independently. Using only a Google Pixel phone and headphones, Project Guideline leverages on-device machine learning (ML) to navigate users along outdoor paths marked with a painted line. The technology has been tested all over the world and even demonstrated during the opening ceremony at the Tokyo 2020 Paralympic Games.

Since the original announcement, we set out to improve Project Guideline by embedding new features, such as obstacle detection and advanced path planning, to safely and reliably navigate users through more complex scenarios (such as sharp turns and nearby pedestrians). The early version featured a simple frame-by-frame image segmentation that detected the position of the path line relative to the image frame. This was sufficient for orienting the user to the line, but provided limited information about the surrounding environment. Improving the navigation signals, such as alerts for obstacles and upcoming turns, required a much better understanding and mapping of the users’ environment. To solve these challenges, we built a platform that can be utilized for a variety of spatially-aware applications in the accessibility space and beyond.

Today, we announce the open source release of Project Guideline, making it available for anyone to use to improve upon and build new accessibility experiences. The release includes source code for the core platform, an Android application, pre-trained ML models, and a 3D simulation framework.

System design

The primary use-case is an Android application, however we wanted to be able to run, test, and debug the core logic in a variety of environments in a reproducible way. This led us to design and build the system using C++ for close integration with MediaPipe and other core libraries, while still being able to integrate with Android using the Android NDK.

Under the hood, Project Guideline uses ARCore to estimate the position and orientation of the user as they navigate the course. A segmentation model, built on the DeepLabV3+ framework, processes each camera frame to generate a binary mask of the guideline (see the previous blog post for more details). Points on the segmented guideline are then projected from image-space coordinates onto a world-space ground plane using the camera pose and lens parameters (intrinsics) provided by ARCore. Since each frame contributes a different view of the line, the world-space points are aggregated over multiple frames to build a virtual mapping of the real-world guideline. The system performs piecewise curve approximation of the guideline world-space coordinates to build a spatio-temporally consistent trajectory. This allows refinement of the estimated line as the user progresses along the path.

Project Guideline builds a 2D map of the guideline, aggregating detected points in each frame (red) to build a stateful representation (blue) as the runner progresses along the path.

A control system dynamically selects a target point on the line some distance ahead based on the user’s current position, velocity, and direction. An audio feedback signal is then given to the user to adjust their heading to coincide with the upcoming line segment. By using the runner’s velocity vector instead of camera orientation to compute the navigation signal, we eliminate noise caused by irregular camera movements common during running. We can even navigate the user back to the line while it’s out of camera view, for example if the user overshot a turn. This is possible because ARCore continues to track the pose of the camera, which can be compared to the stateful line map inferred from previous camera images.

Project Guideline also includes obstacle detection and avoidance features. An ML model is used to estimate depth from single images. To train this monocular depth model, we used SANPO, a large dataset of outdoor imagery from urban, park, and suburban environments that was curated in-house. The model is capable of detecting the depth of various obstacles, including people, vehicles, posts, and more. The depth maps are converted into 3D point clouds, similar to the line segmentation process, and used to detect the presence of obstacles along the user’s path and then alert the user through an audio signal.

Using a monocular depth ML model, Project Guideline constructs a 3D point cloud of the environment to detect and alert the user of potential obstacles along the path.

A low-latency audio system based on the AAudio API was implemented to provide the navigational sounds and cues to the user. Several sound packs are available in Project Guideline, including a spatial sound implementation using the Resonance Audio API. The sound packs were developed by a team of sound researchers and engineers at Google who designed and tested many different sound models. The sounds use a combination of panning, pitch, and spatialization to guide the user along the line. For example, a user veering to the right may hear a beeping sound in the left ear to indicate the line is to the left, with increasing frequency for a larger course correction. If the user veers further, a high-pitched warning sound may be heard to indicate the edge of the path is approaching. In addition, a clear “stop” audio cue is always available in the event the user veers too far from the line, an anomaly is detected, or the system fails to provide a navigational signal.

Project Guideline has been built specifically for Google Pixel phones with the Google Tensor chip. The Google Tensor chip enables the optimized ML models to run on-device with higher performance and lower power consumption. This is critical for providing real-time navigation instructions to the user with minimal delay. On a Pixel 8 there is a 28x latency improvement when running the depth model on the Tensor Processing Unit (TPU) instead of CPU, and 9x improvement compared to GPU.

Testing and simulation

Project Guideline includes a simulator that enables rapid testing and prototyping of the system in a virtual environment. Everything from the ML models to the audio feedback system runs natively within the simulator, giving the full Project Guideline experience without needing all the hardware and physical environment set up.

Screenshot of Project Guideline simulator.

Future direction

To launch the technology forward, WearWorks has become an early adopter and teamed up with Project Guideline to integrate their patented haptic navigation experience, utilizing haptic feedback in addition to sound to guide runners. WearWorks has been developing haptics for over 8 years, and previously empowered the first blind marathon runner to complete the NYC Marathon without sighted assistance. We hope that integrations like these will lead to new innovations and make the world a more accessible place.

The Project Guideline team is also working towards removing the painted line completely, using the latest advancements in mobile ML technology, such as the ARCore Scene Semantics API, which can identify sidewalks, buildings, and other objects in outdoor scenes. We invite the accessibility community to build upon and improve this technology while exploring new use cases in other fields.


Many people were involved in the development of Project Guideline and the technologies behind it. We’d like to thank Project Guideline team members: Dror Avalon, Phil Bayer, Ryan Burke, Lori Dooley, Song Chun Fan, Matt Hall, Amélie Jean-aimée, Dave Hawkey, Amit Pitaru, Alvin Shi, Mikhail Sirotenko, Sagar Waghmare, John Watkinson, Kimberly Wilber, Matthew Willson, Xuan Yang, Mark Zarich, Steven Clark, Jim Coursey, Josh Ellis, Tom Hoddes, Dick Lyon, Chris Mitchell, Satoru Arao, Yoojin Chung, Joe Fry, Kazuto Furuichi, Ikumi Kobayashi, Kathy Maruyama, Minh Nguyen, Alto Okamura, Yosuke Suzuki, and Bryan Tanaka. Thanks to ARCore contributors: Ryan DuToit, Abhishek Kar, and Eric Turner. Thanks to Alec Go, Jing Li, Liviu Panait, Stefano Pellegrini, Abdullah Rashwan, Lu Wang, Qifei Wang, and Fan Yang for providing ML platform support. We’d also like to thank Hartwig Adam, Tomas Izo, Rahul Sukthankar, Blaise Aguera y Arcas, and Huisheng Wang for their leadership support. Special thanks to our partners Guiding Eyes for the Blind and Achilles International.

Source: Google AI Blog

Emerging practices for Society-Centered AI

The first of Google’s AI Principles is to “Be socially beneficial.” As AI practitioners, we’re inspired by the transformative potential of AI technologies to benefit society and our shared environment at a scale and swiftness that wasn’t possible before. From helping address the climate crisis to helping transform healthcare, to making the digital world more accessible, our goal is to apply AI responsibly to be helpful to more people around the globe. Achieving global scale requires researchers and communities to think ahead — and act — collectively across the AI ecosystem.

We call this approach Society-Centered AI. It is both an extension and an expansion of Human-Centered AI, focusing on the aggregate needs of society that are still informed by the needs of individual users, specifically within the context of the larger, shared human experience. Recent AI advances offer unprecedented, societal-level capabilities, and we can now methodically address those needs — if we apply collective, multi-disciplinary AI research to society-level, shared challenges, from forecasting hunger to predicting diseases to improving productivity.

The opportunity for AI to benefit society increases each day. We took a look at our work in these areas and at the research projects we have supported. Recently, Google announced that 70 professors were selected for the 2023 Award for Inclusion Research Program, which supports academic research that addresses the needs of historically marginalized groups globally. Through evaluation of this work, we identified a few emerging practices for Society-Centered AI:

  • Understand society’s needs
    Listening to communities and partners is crucial to understanding major issues deeply and identifying priority challenges to address. As an emerging general purpose technology, AI has the potential to address major global societal issues that can significantly impact people’s lives (e.g., educating workers, improving healthcare, and improving productivity). We have found the key to impact is to be centered on society’s needs. For this, we focus our efforts on goals society has agreed should be prioritized, such as the United Nations’ 17 Sustainable Development Goals, a set of interconnected goals jointly developed by more than 190 countries to address global challenges.
  • Collective efforts to address those needs
    Collective efforts bring stakeholders (e.g., local and academic communities, NGOs, private-public collaborations) into a joint process of design, development, implementation, and evaluation of AI technologies as they are being developed and deployed to address societal needs.
  • Measuring success by how well the effort addresses society’s needs
    It is important and challenging to measure how well AI solutions address society’s needs. In each of our cases, we identified primary and secondary indicators of impact that we optimized through our collaborations with stakeholders.

Why is Society-Centered AI important?

The case examples described below show how the Society-Centered AI approach has led to impact across topics, such as accessibility, health, and climate.

Understanding the needs of individuals with non-standard speech

There are millions of people with non-standard speech (e.g., impaired articulation, dysarthria, dysphonia) in the United States alone. In 2019, Google Research launched Project Euphonia, a methodology that allows individual users with non-standard speech to train personalized speech recognition models. Our success began with the impact we had on each individual who is now able to use voice dictation on their mobile device.

Euphonia started with a Society-Centered AI approach, including collective efforts with the non-profit organizations ALS Therapy Development Institute and ALS Residence Initiative to understand the needs of individuals with amyotrophic lateral sclerosis (ALS) and their ability to use automatic speech recognition systems. Later, we developed the world’s largest corpus of non-standard speech recordings, which enabled us to train a Universal Speech Model to better recognize disordered speech by 37% on real conversation word error rate (WER) measurement. This also led to the 2022 collaboration between the University of Illinois Urbana-Champaign, Alphabet, Apple, Meta, Microsoft, and Amazon to begin the Speech Accessibility Project, an ongoing initiative to create a publicly available dataset of disordered speech samples to improve products and make speech recognition more inclusive of diverse speech patterns. Other technologies that use AI to help remove barriers of modality and languages, include live transcribe, live caption and read aloud.

Focusing on society’s health needs

Access to timely maternal health information can save lives globally: every two minutes a woman dies during pregnancy or childbirth and 1 in 26 children die before reaching age five. In rural India, the education of expectant and new mothers around key health issues pertaining to pregnancy and infancy required scalable, low-cost technology solutions. Together with ARMMAN, Google Research supported a program that uses mobile messaging and machine learning (ML) algorithms to predict when women might benefit from receiving interventions (i.e., targeted preventative care information) and encourages them to engage with the mMitra free voice call program. Within a year, the mMitra program has shown a 17% increase in infants with tripled birth weight and a 36% increase in women understanding the importance of taking iron tablets during pregnancy. Over 175K mothers and growing have been reached through this automated solution, which public health workers use to improve the quality of information delivery.

These efforts have been successful in improving health due to the close collective partnership among the community and those building the AI technology. We have adopted this same approach via collaborations with caregivers to address a variety of medical needs. Some examples include: the use of the Automated Retinal Disease Assessment (ARDA) to help screen for diabetic retinopathy in 250,000 patients in clinics around the world; our partnership with iCAD to bring our mammography AI models to clinical settings to aid in breast cancer detection; and the development of Med-PaLM 2, a medical large language model that is now being tested with Cloud partners to help doctors provide better patient care.

Compounding impact from sustained efforts for crisis response

Google Research’s flood prediction efforts began in 2018 with flood forecasting in India and expanded to Bangladesh to help combat the catastrophic damage from yearly floods. The initial efforts began with partnerships with India’s Central Water Commission, local governments and communities. The implementation of these efforts used SOS Alerts on Search and Maps, and, more recently, broadly expanded access via Flood Hub. Continued collaborations and advancing an AI-based global flood forecasting model allowed us to expand this capability to over 80 countries across Africa, the Asia-Pacific region, Europe, and South, Central, and North America. We also partnered with networks of community volunteers to further amplify flood alerts. By working with governments and communities to measure the impact of these efforts on society, we refined our approach and algorithms each year.

We were able to leverage those methodologies and some of the underlying technology, such as SOS Alerts, from flood forecasting to similar societal needs, such as wildfire forecasting and heat alerts. Our continued engagements with organizations led to the support of additional efforts, such as the World Meteorological Organization's (WMO) Early Warnings For All Initiative. The continued engagement with communities has allowed us to learn about our users' needs on a societal level over time, expand our efforts, and compound the societal reach and impact of our efforts.

Further supporting Society-Centered AI research

We recently funded 18 university research proposals exemplifying a Society-Centered AI approach, a new track within the Google Award for Inclusion Research Program. These researchers are taking the Society-Centered AI methodology and helping create beneficial applications across the world. Examples of some of the projects funded include:

  • AI-Driven Monitoring of Attitude Polarization in Conflict-Affected Countries for Inclusive Peace Process and Women’s Empowerment: This project’s goal is to create LLM-powered tools that can be used to monitor peace in online conversations in developing nations. The initial target communities are where peace is in flux and the effort will put a particular emphasis on mitigating polarization that impacts women and promoting harmony.
  • AI-Assisted Distributed Collaborative Indoor Pollution Meters: A Case Study, Requirement Analysis, and Low-Cost Healthy Home Solution for Indian Communities: This project is looking at the usage of low-cost pollution monitors combined with AI-assisted methodology for identifying recommendations for communities to improve air quality and at home health. The initial target communities are highly impacted by pollution, and the joint work with them includes the goal of developing how to measure improvement in outcomes in the local community.
  • Collaborative Development of AI Solutions for Scaling Up Adolescent Access to Sexual and Reproductive Health Education and Services in Uganda: This project’s goal is to create LLM-powered tools to provide personalized coaching and learning for users' needs on topics of sexual and reproductive health education in low-income settings in Sub-Saharan Africa. The local societal need is significant, with an estimated 25% rate of teenage pregnancy, and the project aims to address the needs with a collective development process for the AI solution.

Future direction

Focusing on society’s needs, working via multidisciplinary collective research, and measuring the impact on society helps lead to AI solutions that are relevant, long-lasting, empowering, and beneficial. See the AI for the Global Goals to learn more about potential Society-Centered AI research problems. Our efforts with non-profits in these areas is complementary to the research that we are doing and encouraging. We believe that further initiatives using Society-Centered AI will help the collective research community solve problems and positively impact society at large.


Many thanks to the many individuals who have worked on these projects at Google including Shruti Sheth, Reena Jana, Amy Chung-Yu Chou, Elizabeth Adkison, Sophie Allweis, Dan Altman, Eve Andersson, Ayelet Benjamini, Julie Cattiau, Yuval Carny, Richard Cave, Katherine Chou, Greg Corrado, Carlos De Segovia, Remi Denton, Dotan Emanuel, Ashley Gardner, Oren Gilon, Taylor Goddu, Brigitte Hoyer Gosselink, Jordan Green, Alon Harris, Avinatan Hassidim, Rus Heywood, Sunny Jansen, Pan-Pan Jiang, Anton Kast, Marilyn Ladewig, Ronit Levavi Morad, Bob MacDonald, Alicia Martin, Shakir Mohamed, Philip Nelson, Moriah Royz, Katie Seaver, Joel Shor, Milind Tambe, Aparna Taneja, Divy Thakkar, Jimmy Tobin, Katrin Tomanek, Blake Walsh, Gal Weiss, Kasumi Widner, Lihong Xi, and teams.

Source: Google AI Blog

Looking back at wildfire research in 2023

Wildfires are becoming larger and affecting more and more communities around the world, often resulting in large-scale devastation. Just this year, communities have experienced catastrophic wildfires in Greece, Maui, and Canada to name a few. While the underlying causes leading to such an increase are complex — including changing climate patterns, forest management practices, land use development policies and many more — it is clear that the advancement of technologies can help to address the new challenges.

At Google Research, we’ve been investing in a number of climate adaptation efforts, including the application of machine learning (ML) to aid in wildfire prevention and provide information to people during these events. For example, to help map fire boundaries, our wildfire boundary tracker uses ML models and satellite imagery to map large fires in near real-time with updates every 15 minutes. To advance our various research efforts, we are partnering with wildfire experts and government agencies around the world.

Today we are excited to share more about our ongoing collaboration with the US Forest Service (USFS) to advance fire modeling tools and fire spread prediction algorithms. Starting from the newly developed USFS wildfire behavior model, we use ML to significantly reduce computation times, thus enabling the model to be employed in near real time. This new model is also capable of incorporating localized fuel characteristics, such as fuel type and distribution, in its predictions. Finally, we describe an early version of our new high-fidelity 3D fire spread model.

Current state of the art in wildfire modeling

Today’s most widely used state-of-the-art fire behavior models for fire operation and training are based on the Rothermel fire model developed at the US Forest Service Fire Lab, by Rothermel et al., in the 1970s. This model considers many key factors that affect fire spread, such as the influence of wind, the slope of the terrain, the moisture level, the fuel load (e.g., the density of the combustible materials in the forest), etc., and provided a good balance between computational feasibility and accuracy at the time. The Rothermel model has gained widespread use throughout the fire management community across the world.

Various operational tools that employ the Rothermel model, such as BEHAVE, FARSITE, FSPro, and FlamMap, have been developed and improved over the years. These tools and the underlying model are used mainly in three important ways: (1) for training firefighters and fire managers to develop their insights and intuitions on fire behavior, (2) for fire behavior analysts to predict the development of a fire during a fire operation and to generate guidance for situation awareness and resource allocation planning, and (3) for analyzing forest management options intended to mitigate fire hazards across large landscapes.  These models are the foundation of fire operation safety and efficiency today.

However, there are limitations on these state-of-the art models, mostly associated with the simplification of the underlying physical processes (which was necessary when these models were created). By simplifying the physics to produce steady state predictions, the required inputs for fuel sources and weather became practical but also more abstract compared to measurable quantities.  As a result, these models are typically “adjusted” and “tweaked” by experienced fire behavior analysts so they work more accurately in certain situations and to compensate for uncertainties and unknowable environmental characteristics. Yet these expert adjustments mean that many of the calculations are not repeatable.

To overcome these limitations, USFS researchers have been working on a new model to drastically improve the physical fidelity of fire behavior prediction. This effort represents the first major shift in fire modeling in the past 50 years. While the new model continues to improve in capturing fire behavior, the computational cost and inference time makes it impractical to be deployed in the field or for applications with near real-time requirements. In a realistic scenario, to make this model useful and practical in training and operations, a speed up of at least 1000x would be needed.

Machine learning acceleration

In partnership with the USFS, we have undertaken a program to apply ML to decrease computation times for complex fire models. Researchers knew that many complex inputs and features could be characterized using a deep neural network, and if successful, the trained model would lower the computational cost and latency of evaluating new scenarios. Deep learning is a branch of machine learning that uses neural networks with multiple hidden layers of nodes that do not directly correspond to actual observations. The model’s hidden layers allow a rich representation of extremely complex systems — an ideal technique for modeling wildfire spread.

We used the USFS physics-based, numerical prediction models to generate many simulations of wildfire behavior and then used these simulated examples to train the deep learning model on the inputs and features to best capture the system behavior accurately. We found that the deep learning model can perform at a much lower computational cost compared to the original and is able to address behaviors resulting from fine-scale processes. In some cases, computation time for capturing the fine-scale features described above and providing a fire spread estimate was 100,000 times faster than running the physics-based numerical models.

This project has continued to make great progress since the first report at presentation at ICFFR 2022 and the USFS Fire Lab's project page provides a glimpse into the ongoing work in this direction. Our team has expanded the dataset used for training by an order of magnitude, from 40M up to 550M training examples. Additionally, we have delivered a prototype ML model that our USFS Fire Lab partner is integrating into a training app that is currently being developed for release in 2024.

Google researchers visiting the USFS Fire Lab in Missoula, MT, stopping by Big Knife Fire Operation Command Center.

Fine-grained fuel representation

Besides training, another key use-case of the new model is for operational fire prediction. To fully leverage the advantages of the new model’s capability to capture the detailed fire behavior changes from small-scale differences in fuel structures, high resolution fuel mapping and representation are needed. To this end, we are currently working on the integration of high resolution satellite imagery and geo information into ML models to allow fuel specific mapping at-scale. Some of the preliminary results will be presented at the upcoming 10th International Fire Ecology and Management Congress in November 2023.

Future work

Beyond the collaboration on the new fire spread model, there are many important and challenging problems that can help fire management and safety. Many such problems require even more accurate fire models that fully consider 3D flow interactions and fluid dynamics, thermodynamics and combustion physics. Such detailed calculations usually require high-performance computers (HPCs) or supercomputers.

These models can be used for research and longer-term planning purposes to develop insights on extreme fire development scenarios, build ML classification models, or establish a meaningful “danger index” using the simulated results. These high-fidelity simulations can also be used to supplement physical experiments that are used in expanding the operational models mentioned above.

In this direction, Google research has also developed a high-fidelity large-scale 3D fire simulator that can be run on Google TPUs. In the near future, there is a plan to further leverage this new capability to augment the experiments, and to generate data to build insights on the development of extreme fires and use the data to design a fire-danger classifier and fire-danger index protocol.

An example of 3D high-fidelity simulation. This is a controlled burn field experiment (FireFlux II) simulated using Google’s high fidelity fire simulator.


We thank Mark Finney, Jason Forthofer, William Chatham and Issac Grenfell from US Forest Service Missoula Fire Science Laboratory and our colleagues John Burge, Lily Hu, Qing Wang, Cenk Gazen, Matthias Ihme, Vivian Yang, Fei Sha and John Anderson for core contributions and useful discussions. We also thank Tyler Russell for his assistance with program management and coordination.

Source: Google AI Blog

Improving traffic evacuations: A case study

Some cities or communities develop an evacuation plan to be used in case of an emergency. There are a number of reasons why city officials might enact their plan, a primary one being a natural disaster, such as a tornado, flood, or wildfire. An evacuation plan can help the community more effectively respond to an emergency, and so could help save lives. However, it can be difficult for a city to evaluate such a plan because it is not practical to have an entire town or city rehearse a full blown evacuation. For example, Mill Valley, a city in northern California, created a wildfire evacuation plan but lacked an estimate for how long the evacuation would take.

Today we describe a case study in which we teamed up with the city of Mill Valley to test and improve their evacuation plan. We outline our approach in our paper, “Mill Valley Evacuation Study”. We started by using a traffic simulator to model a citywide evacuation. The research goal was to provide the city with detailed estimates for how long it would take to evacuate the city, and, by studying the egress pattern, to find modifications to make the plan more effective. While our prior work on this subject provided an estimate for the evacuation time and showed how the time could be reduced if certain road changes were implemented, it turns out the recommendations in that paper — such as changing the number of outgoing lanes on an arterial — were not feasible. The current round of research improves upon the initial study by more accurately modeling the number and starting locations of vehicles, by using a more realistic map, and by working closely with city officials to ensure that recommended changes to the plan are deemed viable.

Geography and methodology

Mill Valley is in Marin County, California, north of San Francisco. Many of the residences are located on the steep hillsides of several valleys surrounded by dense redwood forests.

Aerial views of Mill Valley, courtesy of the City of Mill Valley.

Many of those residences are in areas that have only one exit direction, toward the town center. From there the best evacuation route is toward Highway 101, which is in the flat part of the city and is the most likely area to be far from potential wildfires. Some neighborhoods have other routes that lead away from both the city and Highway 101, but those routes pass through hilly forested areas, which could be dangerous or impassable during a wildfire. So, the evacuation plan directs all vehicles west of Highway 101 to head east, to the highway (see map below). The neighborhoods east of Highway 101 are not included in the simulation because they are away from areas with a high fire hazard rating, and are close to the highway.

Mill Valley has about 11,400 households west of Highway 101. Most Mill Valley households have two vehicles. Evacuation times scale with the number of vehicles, so it is in the common interest to minimize the number of vehicles used during an evacuation. To that end, Mill Valley has a public awareness campaign aimed at having each household evacuate in one vehicle. While no one knows how many vehicles would be used during an evacuation, it is safe to assume it is on average between one and two per household. The basic evacuation problem, then, is how to efficiently get between 11 and 23 thousand vehicles from the various residences onto one of the three sets of Highway 101 on-ramps.

The simulated part of Mill Valley west of Highway 101 is inside the blue border. Highway 101 is shown in green. The red squares indicate the three sets of Highway 101 on-ramps. The pink area has the highest fire hazard rating.

The current work uses the same general methodology as the previous research, namely, running the open source SUMO agent-based traffic simulator on a map of Mill Valley. The traffic simulator models traffic by simulating each vehicle individually. The detailed behaviors of vehicles are dictated by a car-following model. Each vehicle is given a point and time at which to start and an initial route. The routes of most vehicles are updated throughout the simulation, depending on conditions. To consider potential changes in driver behavior under the high stress conditions of an evacuation, the effects of the “aggressiveness” of each car is also investigated, but in our case the impacts are minimal. Some simplifying assumptions are that vehicles originate at residential addresses and the roads and highways are initially empty. These assumptions correspond approximately to conditions that could be encountered if an evacuation happens in the middle of the night. The main inputs in the simulation are the road network, the household locations, the average number of vehicles per household, and a departure temporal distribution. We have to make assumptions about the departure distribution. After discussing with the city officials, we chose a distribution such that most vehicles depart within an hour.

Four bottlenecks

Mill Valley has three sets of Highway 101 on-ramps: northern, middle, and southern. All the vehicles must use one of these sets of on-ramps to reach their destination (either the northernmost or southernmost segment of Highway 101 included in our map). Given that we are only concerned with the majority of Mill Valley that lies west of the highway, there are two lanes that approach the northern on-ramps, and one lane that approaches each of the middle and southern on-ramps. Since every vehicle has to pass over one of these four lanes to reach the highway, they are the bottlenecks. Given the geography and existing infrastructure, adding more lanes is infeasible. The aim of this research, then, is to try to modify traffic patterns to maximize the rate of traffic on each of the four lanes.

Evacuation plan

When we started this research, Mill Valley had a preliminary evacuation plan. It included modifying traffic patterns — disabling traffic lights and changing traffic rules — on a few road segments, as well as specifying the resources (traffic officers, signage) necessary to implement the changes. As an example, a two-way road may be changed to a one-way road to double the number of outgoing lanes. Temporarily changing the direction of traffic is called contraflow.

The plot below shows the simulated fraction of vehicles that have departed or reached their destinations versus time, for 1, 1.5, and 2 vehicles per household (left to right). The dashed line on the far left shows the fraction that have departed. The solid black lines show the preliminary evacuation plan results and the dotted lines indicate the normal road network (baseline) results. The preliminary evacuation plan significantly speeds up the evacuation.

The cumulative fraction of vehicles vs. time in hours. The demand curve is shown in the dashed line on the far left. The solid lines show the preliminary evacuation plan curves for 1, 1.5 and 2 vehicles per household (left to right). The dotted lines show the same for the baseline case.

We can understand how effective the preliminary evacuation plan is by measuring the rates at the bottlenecks. The below plots show the rate of traffic on each of the four lanes leading to the highway on-ramps for the case of 1.5 vehicles per household for both the baseline case (the normal road rules; shown shaded in gray) and the preliminary evacuation plan (shown outlined in black). The average rate per lane varies greatly in the different cases. It is clear that, while the evacuation plan leads to increased evacuation rates, there is room for improvement. In particular, the middle on-ramps are quite underutilized.

The rates of traffic on the four lanes leading to Highway 101 on-ramps for both the baseline case (normal road rules; shown shaded in gray) and the preliminary evacuation plan (shown outlined in black).

Final evacuation plan

After studying the map and investigating different alternatives, we, working together with city officials, found a minimal set of new road changes that substantially lower the evacuation time compared to the preliminary evacuation plan (shown below). We call this the final evacuation plan. It extends the contraflow section of the preliminary plan 1000 feet further west, to a main intersection. Crucially, this allows for one of the (normally) two outgoing lanes to be dedicated to routing traffic to the middle on-ramps. It also creates two outgoing lanes from that main intersection clear through to the northern on-ramps, over ¾ of a mile to the east.

A map of the main changes in the final evacuation plan. The red line shows that traffic heading north on Camino Alto gets diverted to the middle Highway 101 on-ramps. The blue line shows traffic in the northern lane of E Blithedale Ave gets routed on the new contraflow section.

The rate per lane plots comparing the preliminary and final evacuation plans are shown below for 1.5 vehicles per household. The simulation indicates that the final plan increases the average rate of traffic on the lane leading to the middle on-ramps from about 4 vehicles per minute to about 18. It also increases the through rate of the northern on-ramps by over 60%.

The rates of traffic on the four lanes leading to Highway 101 on-ramps for both the preliminary case (shown shaded in gray) and the final evacuation plan (shown outlined in black).

The below plot shows the cumulative fraction of vehicles vs. time, comparing the cases of 1, 1.5 and 2 vehicles per household for the preliminary and final evacuation plans. The speedup is quite significant, on the scale of hours. For example, with 1.5 vehicles per household, it took 5.3 hours to evacuate the city using the preliminary evacuation plan, and only 3.5 hours using the final plan.

The cumulative fraction of vehicles vs. time in hours. The demand curve is shown in the dashed line on the far left. The solid lines show the final evacuation plan curves for 1, 1.5 and 2 vehicles per household (left to right). The dotted lines show the same for the preliminary evacuation plan.


Evacuation plans can be crucial in quickly getting many people to safety in emergency situations. While some cities have traffic evacuation plans in place, it can be difficult for officials to learn how well the plan works or whether it can be improved. Google Research helped Mill Valley test and evaluate their evacuation plan by running traffic simulations. We found that, while the preliminary plan did speed up the evacuation time, some minor changes to the plan significantly expedited evacuation. We worked closely with the city during this research, and Mill Valley has adopted the final plan. We were able to provide the city with more simulation details, including results for evacuating the city one area at a time. Full details can be found in the paper.

Detailed recommendations for a particular evacuation plan are necessarily specific to the area under study. So, the specific road network changes we found for Mill Valley are not directly applicable for other cities. However, we used only public data (road network from OpenStreetMap; household information from census data) and an open source simulator (SUMO), so any city or agency could use the methodology used in our paper to obtain results for their area.


We thank former Mayor John McCauley and City of Mill Valley personnel Tom Welch, Lindsay Haynes, Danielle Staude, Rick Navarro and Alan Piombo for numerous discussions and feedback, and Carla Bromberg for program management.

Source: Google AI Blog

SANPO: A Scene understanding, Accessibility, Navigation, Pathfinding, & Obstacle avoidance dataset

As most people navigate their everyday world, they process visual input from the environment using an eye-level perspective. Unlike robots and self-driving cars, people don't have any "out-of-body" sensors to help guide them. Instead, a person’s sensory input is completely "egocentric", or "from the self." This also applies to new technologies that understand the world around us from a human-like perspective, e.g., robots navigating through unknown buildings, AR glasses that highlight objects, or assistive technology to help people run independently.

In computer vision, scene understanding is the subfield that studies how visible objects relate to the scene's 3D structure and layout by focusing on the spatial, functional, and semantic relationships between objects and their environment. For example, autonomous drivers must understand the 3D structure of the road, sidewalks, and surrounding buildings while identifying and recognizing street signs and stop lights, a task made easier with 3D data from a special laser scanner mounted on the top of the car rather than 2D images from the driver’s perspective. Robots navigating a park must understand where the path is and what obstacles might interfere, which is simplified with a map of their surroundings and GPS positioning data. Finally, AR glasses that help users find their way need to understand where the user is and what they are looking at.

The computer vision community typically studies scene understanding tasks in contexts like self-driving, where many other sensors (GPS, wheel positioning, maps, etc.) beyond egocentric imagery are available. Yet most datasets in this space do not focus exclusively on egocentric data, so they are less applicable to human-centered navigation tasks. While there are plenty of self-driving focused scene understanding datasets, they have limited generalization to egocentric human scene understanding. A comprehensive human egocentric dataset would help build systems for related applications and serve as a challenging benchmark for the scene understanding community.

To that end, we present the Scene understanding, Accessibility, Navigation, Pathfinding, Obstacle avoidance dataset, or SANPO (also the Japanese word for ”brisk stroll”), a multi-attribute video dataset for outdoor human egocentric scene understanding. The dataset consists of real world data and synthetic data, which we call SANPO-Real and SANPO-Synthetic, respectively. It supports a wide variety of dense prediction tasks, is challenging for current models, and includes real and synthetic data with depth maps and video panoptic masks in which each pixel is assigned a semantic class label (and for some semantic classes, each pixel is also assigned a semantic instance ID that uniquely identifies that object in the scene). The real dataset covers diverse environments and has videos from two stereo cameras to support multi-view methods, including 11.4 hours captured at 15 frames per second (FPS) with dense annotations. Researchers can download and use SANPO here.

3D scene of a real session built using the provided annotations (segmentation, depth and camera positions). The top center video shows the depth map, and the top right shows the RGB or semantic annotations.


SANPO-Real is a multiview video dataset containing 701 sessions recorded with two stereo cameras: a head-mounted ZED Mini and a chest-mounted ZED-2i. That’s four RGB streams per session at 15 FPS. 597 sessions are recorded at a resolution of 2208x1242 pixels, and the remainder are recorded at a resolution of 1920x1080 pixels. Each session is approximately 30 seconds long, and the recorded videos are rectified using Zed software and saved in a lossless format. Each session has high-level attribute annotations, camera pose trajectories, dense depth maps from CREStereo, and sparse depth maps provided by the Zed SDK. A subset of sessions have temporally consistent panoptic segmentation annotations of each instance.

The SANPO data collection system for collecting real-world data. Right: (i) a backpack with ZED 2i and ZED Mini cameras for data collection (bottom), (ii) the inside of the backpack showing the ZED box and battery pack mounted on a 3D printed container (middle), and (iii) an Android app showing the live feed from the ZED cameras (top). Left: The chest-mounted ZED-2i has a stereo baseline of 12cm with a 2.1mm focal length, and the head-mounted ZED Mini has a baseline of 6.3cm with a 2.1mm focal length.

Temporally consistent panoptic segmentation annotation protocol

SANPO includes thirty different class labels, including various surfaces (road, sidewalk, curb, etc.), fences (guard rails, walls,, gates), obstacles (poles, bike racks, trees), and creatures (pedestrians, riders, animals). Gathering high-quality annotations for these classes is an enormous challenge. To provide temporally consistent panoptic segmentation annotation we divide each video into 30-second sub-videos and annotate every fifth frame (90 frames per sub-video), using a cascaded annotation protocol. At each stage, we ask annotators to draw borders around five mutually exclusive labels at a time. We send the same image to different annotators with as many stages as it takes to collect masks until all labels are assigned, with annotations from previous subsets frozen and shown to the annotator. We use AOT, a machine learning model that reduces annotation effort by giving annotators automatic masks from which to start, taken from previous frames during the annotation process. AOT also infers segmentation annotations for intermediate frames using the manually annotated preceding and following frames. Overall, this approach reduces annotation time, improves boundary precision, and ensures temporally consistent annotations for up to 30 seconds.

Temporally consistent panoptic segmentation annotations. The segmentation mask’s title indicates whether it was manually annotated or AOT propagated.


Real-world data has imperfect ground truth labels due to hardware, algorithms, and human mistakes, whereas synthetic data has near-perfect ground truth and can be customized. We partnered with Parallel Domain, a company specializing in lifelike synthetic data generation, to create SANPO-Synthetic, a high-quality synthetic dataset to supplement SANPO-Real. Parallel Domain is skilled at creating handcrafted synthetic environments and data for machine learning applications. Thanks to their work, SANPO-Synthetic matches real-world capture conditions with camera parameters, placement, and scenery.

3D scene of a synthetic session built using the provided annotations (segmentation, depth and odometry). The top center video shows the depth map, and the top right shows the RGB or semantic annotations.

SANPO-Synthetic is a high quality video dataset, handcrafted to match real world scenarios. It contains 1961 sessions recorded using virtualized Zed cameras, evenly split between chest-mounted and head-mounted positions and calibrations. These videos are monocular, recorded from the left lens only. These sessions vary in length and FPS (5, 14.28, and 33.33) for a mix of temporal resolution / length tradeoffs, and are saved in a lossless format. All the sessions have precise camera pose trajectories, dense pixel accurate depth maps and temporally consistent panoptic segmentation masks.

SANPO-Synthetic data has pixel-perfect annotations, even for small and distant instances. This helps develop challenging datasets that mimic the complexity of real-world scenes. SANPO-Synthetic and SANPO-Real are also drop-in replacements for each other, so researchers can study domain transfer tasks or use synthetic data during training with few domain-specific assumptions.

An even sampling of real and synthetic scenes.


Semantic classes

We designed our SANPO taxonomy: i) with human egocentric navigation in mind, ii) with the goal of being reasonably easy to annotate, and iii) to be as close as possible to the existing segmentation taxonomies. Though built with human egocentric navigation in mind, it can be easily mapped or extended to other human egocentric scene understanding applications. Both SANPO-Real and SANPO-Synthetic feature a wide variety of objects one would expect in egocentric obstacle detection data, such as roads, buildings, fences, and trees. SANPO-Synthetic includes a broad distribution of hand-modeled objects, while SANPO-Real features more “long-tailed” classes that appear infrequently in images, such as gates, bus stops, or animals.

Distribution of images across the classes in the SANPO taxonomy.

Instance masks

SANPO-Synthetic and a portion of SANPO-Real are also annotated with panoptic instance masks, which assign each pixel to a class and instance ID. Because it is generally human-labeled, SANPO-Real has a large number of frames with generally less than 20 instances per frame. Similarly, SANPO-Synthetic’s virtual environment offers pixel-accurate segmentation of most unique objects in the scene. This means that synthetic images frequently feature many more instances within each frame.

When considering per-frame instance counts, synthetic data frequently features many more instances per frame than the labeled portions of SANPO-Real.

Comparison to other datasets

We compare SANPO to other important video datasets in this field, including SCAND, MuSoHu, Ego4D, VIPSeg, and Waymo Open. Some of these are intended for robot navigation (SCAND) or autonomous driving (Waymo) tasks. Across these datasets, only Waymo Open and SANPO have both panoptic segmentations and depth maps, and only SANPO has both real and synthetic data.

Comparison to other video datasets. For stereo vs mono video, datasets marked with ★ have stereo video for all scenes and those marked ☆ provide stereo video for a subset. For depth maps, ★ indicates dense depth while ☆ represents sparse depth, e.g., from a lower-resolution LIDAR scanner.

Conclusion and future work

We present SANPO, a large-scale and challenging video dataset for human egocentric scene understanding, which includes real and synthetic samples with dense prediction annotations. We hope SANPO will help researchers build visual navigation systems for the visually impaired and advance visual scene understanding. Additional details are available in the preprint and on the SANPO dataset GitHub repository.


This dataset was the outcome of hard work of many individuals from various teams within Google and our external partner, Parallel Domain.

Core Team: Mikhail Sirotenko, Dave Hawkey, Sagar Waghmare, Kimberly Wilber, Xuan Yang, Matthew Wilson

Parallel Domain: Stuart Park, Alan Doucet, Alex Valence-Lanoue, & Lars Pandikow.

We would also like to thank following team members: Hartwig Adam, Huisheng Wang, Lucian Ionita, Nitesh Bharadwaj, Suqi Liu, Stephanie Debats, Cattalyya Nuengsigkapian, Astuti Sharma, Alina Kuznetsova, Stefano Pellegrini, Yiwen Luo, Lily Pagan, Maxine Deines, Alex Siegman, Maura O’Brien, Rachel Stigler, Bobby Tran, Supinder Tohra, Umesh Vashisht, Sudhindra Kopalle, Reet Bhatia.

Source: Google AI Blog

Scalable spherical CNNs for scientific applications

Typical deep learning models for computer vision, like convolutional neural networks (CNNs) and vision transformers (ViT), process signals assuming planar (flat) spaces. For example, digital images are represented as a grid of pixels on a plane. However, this type of data makes up only a fraction of the data we encounter in scientific applications. Variables sampled from the Earth's atmosphere, like temperature and humidity, are naturally represented on the sphere. Some kinds of cosmological data and panoramic photos are also spherical signals, and are better treated as such.

Using methods designed for planar images to process spherical signals is problematic for a couple of reasons. First, there is a sampling problem, i.e., there is no way of defining uniform grids on the sphere, which are needed for planar CNNs and ViTs, without heavy distortion.

When projecting the sphere into a plane, the patch represented by the red circle is heavily distorted near the poles. This sampling problem hurts the accuracy of conventional CNNs and ViTs on spherical inputs.

Second, signals and local patterns on the sphere are often complicated by rotations, so models need a way to address that. We would like equivariance to 3D rotations, which ensures that learned features follow the rotations of the input. This leads to better utilization of the model parameters and allows training with less data. Equivariance to 3D rotations is also useful in most settings where inputs don’t have a preferred orientation, such as 3D shapes and molecules.

Drone racing with panoramic cameras. Here the sharp turns result in large 3D rotations of the spherical image. We would like our models to be robust to such rotations. Source: https://www.youtube.com/watch?v=_J7qXbbXY80 (licensed under CC BY)
In the atmosphere, it is common to see similar patterns appearing at different positions and orientations. We would like our models to share parameters to recognize these patterns.

With the above challenges in mind, in “Scaling Spherical CNNs”, presented at ICML 2023, we introduce an open-source library in JAX for deep learning on spherical surfaces. We demonstrate how applications of this library match or surpass state-of-the-art performance on weather forecasting and molecular property prediction benchmarks, tasks that are typically addressed with transformers and graph neural networks.

Background on spherical CNNs

Spherical CNNs solve both the problems of sampling and of robustness to rotation by leveraging spherical convolution and cross-correlation operations, which are typically computed via generalized Fourier transforms. For planar surfaces, however, convolution with small filters is faster, because it can be performed on regular grids without using Fourier transforms. The higher computational cost for spherical inputs has so far restricted the application of spherical CNNs to small models and datasets and low resolution datasets.

Our contributions

We have implemented the spherical convolutions from spin-weighted spherical CNNs in JAX with a focus on speed, and have enabled distributed training over a large number of TPUs using data parallelism. We also introduced a new phase collapse activation and spectral batch normalization layer, and a new residual block that improves accuracy and efficiency, which allows training more accurate models up to 100x larger than before. We apply these new models on molecular property regression and weather forecasting.

We scale spherical CNNs by up to two orders of magnitude in terms of feature sizes and model capacity, compared to the literature: Cohen’18Esteves’18Esteves’20, and Cobb’21VGG-19 is included as a conventional CNN reference. Our largest model for weather forecasting has 256 x 256 x 78 inputs and outputs, and runs 96 convolutional layers during training with a lowest internal resolution of 128 x 128 x 256.

Molecular property regression

Predicting properties of molecules has applications in drug discovery, where the goal is to quickly screen numerous molecules in search of those with desirable properties. Similar models may also be relevant in the design of drugs targeting the interaction between proteins. Current methods in computational or experimental quantum chemistry are expensive, which motivates the use of machine learning.

Molecules can be represented by a set of atoms and their positions in 3D space; rotations of the molecule change the positions but not the molecular properties. This motivates the application of spherical CNNs because of their rotation equivariance. However, molecules are not defined as signals on the sphere so the first step is to map them to a set of spherical functions. We do so by leveraging physics-based interactions between the atoms of the molecule.

Each atom is represented by a set of spherical signals accumulating physical interactions with other atoms of each type (shown in the three panels on the right). For example, the oxygen atom (O; top panel) has a channel for oxygen (indicated by the sphere labeled “O” on the left) and hydrogen (“H”, right). The accumulated Coulomb forces on the oxygen atom with respect to the two hydrogen atoms is indicated by the red shaded regions on the bottom of the sphere labeled “H”. Because the oxygen atom contributes no forces to itself, the “O” sphere is uniform. We include extra channels for the Van der Waals forces.

Spherical CNNs are applied to each atom's features, and results are later combined to produce the property predictions. This results in state-of-the art performance in most properties as typically evaluated in the QM9 benchmark:

Error comparison against the state-of-the-art on 12 properties of QM9 (see the dataset paper for details). We show TorchMD-Net and PaiNN results, normalizing TorchMD-Net errors to 1.0 (lower is better). Our model, shown in green, outperforms the baselines in most targets.

Weather forecasting

Accurate climate forecasts serve as invaluable tools for providing timely warnings of extreme weather events, enabling effective water resource management, and guiding informed infrastructure planning. In a world increasingly threatened by climate disasters, there is an urgency to deliver forecasts much faster and more accurately over a longer time horizon than general circulation models. Forecasting models will also be important for predicting the safety and effectiveness of efforts intended to combat climate change, such as climate interventions. The current state-of-the-art uses costly numerical models based on fluid dynamics and thermodynamics, which tend to drift after a few days.

Given these challenges, there is an urgency for machine learning researchers to address climate forecasting problems, as data-driven techniques have the potential of both reducing the computational cost and improving long range accuracy. Spherical CNNs are suitable for this task since atmospheric data is natively presented on the sphere. They can also efficiently handle repeating patterns at different positions and orientations that are common in such data.

We apply our models to several weather forecasting benchmarks and outperform or match neural weather models based on conventional CNNs (specifically, 1, 2, and 3). Below we show results in a test setting where the model takes a number of atmospheric variables as input and predicts their values six hours ahead. The model is then iteratively applied on its own predictions to produce longer forecasts. During training, the model predicts up to three days ahead, and is evaluated up to five days. Keisler proposed a graph neural network for this task, but we show that spherical CNNs can match the GNN accuracy in the same setting.

Iterative weather forecasting up to five days (120h) ahead with spherical CNNs. The animations show the specific humidity forecast at a given pressure and its error.
Wind speed and temperature forecasts with spherical CNNs.

Additional resources

Our JAX library for efficient spherical CNNs is now available. We have shown applications to molecular property regression and weather forecasting, and we believe the library will be helpful in other scientific applications, as well as in computer vision and 3D vision.

Weather forecasting is an active area of research at Google with the goal of building more accurate and robust models — like Graphcast, a recent ML-based mid-range forecasting model — and to build tools that enable further advancement across the research community, such as the recently released WeatherBench 2.


This work was done in collaboration with Jean-Jacques Slotine, and is based on previous collaborations with Kostas Daniilidis and Christine Allen-Blanchette. We thank Stephan Hoyer, Stephan Rasp, and Ignacio Lopez-Gomez for helping with data processing and evaluation, and Fei Sha, Vivian Yang, Anudhyan Boral, Leonardo Zepeda-Núñez, and Avram Hershko for suggestions and discussions. We are thankful to Michael Riley and Corinna Cortes for supporting and encouraging this project.

Source: Google AI Blog

Responsible AI at Google Research: AI for Social Good

Google’s AI for Social Good team consists of researchers, engineers, volunteers, and others with a shared focus on positive social impact. Our mission is to demonstrate AI’s societal benefit by enabling real-world value, with projects spanning work in public health, accessibility, crisis response, climate and energy, and nature and society. We believe that the best way to drive positive change in underserved communities is by partnering with change-makers and the organizations they serve.

In this blog post we discuss work done by Project Euphonia, a team within AI for Social Good, that aims to improve automatic speech recognition (ASR) for people with disordered speech. For people with typical speech, an ASR model’s word error rate (WER) can be less than 10%. But for people with disordered speech patterns, such as stuttering, dysarthria and apraxia, the WER could reach 50% or even 90% depending on the etiology and severity. To help address this problem, we worked with more than 1,000 participants to collect over 1,000 hours of disordered speech samples and used the data to show that ASR personalization is a viable avenue for bridging the performance gap for users with disordered speech. We've shown that personalization can be successful with as little as 3-4 minutes of training speech using layer freezing techniques.

This work led to the development of Project Relate for anyone with atypical speech who could benefit from a personalized speech model. Built in partnership with Google’s Speech team, Project Relate enables people who find it hard to be understood by other people and technology to train their own models. People can use these personalized models to communicate more effectively and gain more independence. To make ASR more accessible and usable, we describe how we fine-tuned Google’s Universal Speech Model (USM) to better understand disordered speech out of the box, without personalization, for use with digital assistant technologies, dictation apps, and in conversations.

Addressing the challenges

Working closely with Project Relate users, it became clear that personalized models can be very useful, but for many users, recording dozens or hundreds of examples can be challenging. In addition, the personalized models did not always perform well in freeform conversation.

To address these challenges, Euphonia’s research efforts have been focusing on speaker independent ASR (SI-ASR) to make models work better out of the box for people with disordered speech so that no additional training is necessary.

Prompted Speech dataset for SI-ASR

The first step in building a robust SI-ASR model was to create representative dataset splits. We created the Prompted Speech dataset by splitting the Euphonia corpus into train, validation and test portions, while ensuring that each split spanned a range of speech impairment severity and underlying etiology and that no speakers or phrases appeared in multiple splits. The training portion consists of over 950k speech utterances from over 1,000 speakers with disordered speech. The test set contains around 5,700 utterances from over 350 speakers. Speech-language pathologists manually reviewed all of the utterances in the test set for transcription accuracy and audio quality.

Real Conversation test set

Unprompted or conversational speech differs from prompted speech in several ways. In conversation, people speak faster and enunciate less. They repeat words, repair misspoken words, and use a more expansive vocabulary that is specific and personal to themselves and their community. To improve a model for this use case, we created the Real Conversation test set to benchmark performance.

The Real Conversation test set was created with the help of trusted testers who recorded themselves speaking during conversations. The audio was reviewed, any personally identifiable information (PII) was removed, and then that data was transcribed by speech-language pathologists. The Real Conversation test set contains over 1,500 utterances from 29 speakers.

Adapting USM to disordered speech

We then tuned USM on the training split of the Euphonia Prompted Speech set to improve its performance on disordered speech. Instead of fine-tuning the full model, our tuning was based on residual adapters, a parameter-efficient tuning approach that adds tunable bottleneck layers as residuals between the transformer layers. Only these layers are tuned, while the rest of the model weights are untouched. We have previously shown that this approach works very well to adapt ASR models to disordered speech. Residual adapters were only added to the encoder layers, and the bottleneck dimension was set to 64.


To evaluate the adapted USM, we compared it to older ASR models using the two test sets described above. For each test, we compare adapted USM to the pre-USM model best suited to that task: (1) For short prompted speech, we compare to Google’s production ASR model optimized for short form ASR; (2) for longer Real Conversation speech, we compare to a model trained for long form ASR. USM improvements over pre-USM models can be explained by USM’s relative size increase, 120M to 2B parameters, and other improvements discussed in the USM blog post.

Model word error rates (WER) for each test set (lower is better).

We see that the USM adapted with disordered speech significantly outperforms the other models. The adapted USM’s WER on Real Conversation is 37% better than the pre-USM model, and on the Prompted Speech test set, the adapted USM performs 53% better.

These findings suggest that the adapted USM is significantly more usable for an end user with disordered speech. We can demonstrate this improvement by looking at transcripts of Real Conversation test set recordings from a trusted tester of Euphonia and Project Relate (see below).

Audio1    Ground Truth    Pre-USM ASR    Adapted USM
   I now have an Xbox adaptive controller on my lap.    i now have a lot and that consultant on my mouth    i now had an xbox adapter controller on my lamp.
   I've been talking for quite a while now. Let's see.    quite a while now    i've been talking for quite a while now.
Example audio and transcriptions of a trusted tester’s speech from the Real Conversation test set.

A comparison of the Pre-USM and adapted USM transcripts revealed some key advantages:

  • The first example shows that Adapted USM is better at recognizing disordered speech patterns. The baseline misses key words like “XBox” and “controller” that are important for a listener to understand what they are trying to say.
  • The second example is a good example of how deletions are a primary issue with ASR models that are not trained with disordered speech. Though the baseline model did transcribe a portion correctly, a large part of the utterance was not transcribed, losing the speaker’s intended message.


We believe that this work is an important step towards making speech recognition more accessible to people with disordered speech. We are continuing to work on improving the performance of our models. With the rapid advancements in ASR, we aim to ensure people with disordered speech benefit as well.


Key contributors to this project include Fadi Biadsy, Michael Brenner, Julie Cattiau, Richard Cave, Amy Chung-Yu Chou, Dotan Emanuel, Jordan Green, Rus Heywood, Pan-Pan Jiang, Anton Kast, Marilyn Ladewig, Bob MacDonald, Philip Nelson, Katie Seaver, Joel Shor, Jimmy Tobin, Katrin Tomanek, and Subhashini Venugopalan. We gratefully acknowledge the support Project Euphonia received from members of the USM research team including Yu Zhang, Wei Han, Nanxin Chen, and many others. Most importantly, we wanted to say a huge thank you to the 2,200+ participants who recorded speech samples and the many advocacy groups who helped us connect with these participants.

1Audio volume has been adjusted for ease of listening, but the original files would be more consistent with those used in training and would have pauses, silences, variable volume, etc. 

Source: Google AI Blog

Natural Language Assessment: A New Framework to Promote Education

Whether it's a professional honing their skills or a child learning to read, coaches and educators play a key role in assessing the learner's answer to a question in a given context and guiding them towards a goal. These interactions have unique characteristics that set them apart from other forms of dialogue, yet are not available when learners practice alone at home. In the field of natural language processing, this type of capability has not received much attention and is technologically challenging. We set out to explore how we can use machine learning to assess answers in a way that facilitates learning.

In this blog, we introduce an important natural language understanding (NLU) capability called Natural Language Assessment (NLA), and discuss how it can be helpful in the context of education. While typical NLU tasks focus on the user's intent, NLA allows for the assessment of an answer from multiple perspectives. In situations where a user wants to know how good their answer is, NLA can offer an analysis of how close the answer is to what is expected. In situations where there may not be a “correct” answer, NLA can offer subtle insights that include topicality, relevance, verbosity, and beyond. We formulate the scope of NLA, present a practical model for carrying out topicality NLA, and showcase how NLA has been used to help job seekers practice answering interview questions with Google's new interview prep tool, Interview Warmup.

Overview of Natural Language Assessment (NLA)

The goal of NLA is to evaluate the user's answer against a set of expectations. Consider the following components for an NLA system interacting with students:

  • A question presented to the student
  • Expectations that define what we expect to find in the answer (e.g., a concrete textual answer, a set of topics we expect the answer to cover, conciseness)
  • An answer provided by the student
  • An assessment output (e.g., correctness, missing information, too specific or general, stylistic feedback, pronunciation, etc.)
  • [Optional] A context (e.g., a chapter in a book or an article)

With NLA, both the expectations about the answer and the assessment of the answer can be very broad. This enables teacher-student interactions that are more expressive and subtle. Here are two examples:

  1. A question with a concrete correct answer: Even in situations where there is a clear correct answer, it can be helpful to assess the answer more subtly than simply correct or incorrect. Consider the following:

    Context: Harry Potter and the Philosopher's Stone
    Question: “What is Hogwarts?”
    Expectation: “Hogwarts is a school of Witchcraft and Wizardry” [expectation is given as text]
    Answer: “I am not exactly sure, but I think it is a school.”

    The answer may be missing salient details but labeling it as incorrect wouldn’t be entirely true or useful to a user. NLA can offer a more subtle understanding by, for example, identifying that the student’s answer is too general, and also that the student is uncertain.

    Illustration of the NLA process from input question, answer and expectation to assessment output

    This kind of subtle assessment, along with noting the uncertainty the student expressed, can be important in helping students build skills in conversational settings.

  2. Topicality expectations: There are many situations in which a concrete answer is not expected. For example, if a student is asked an opinion question, there is no concrete textual expectation. Instead, there's an expectation of relevance and opinionation, and perhaps some level of succinctness and fluency. Consider the following interview practice setup:

    Question: “Tell me a little about yourself?”
    Expectations: { “Education”, “Experience”, “Interests” } (a set of topics)
    Answer: “Let’s see. I grew up in the Salinas valley in California and went to Stanford where I majored in economics but then got excited about technology so next I ….”

    In this case, a useful assessment output would map the user’s answer to a subset of the topics covered, possibly along with a markup of which parts of the text relate to which topic. This can be challenging from an NLP perspective as answers can be long, topics can be mixed, and each topic on its own can be multi-faceted.

A Topicality NLA Model

In principle, topicality NLA is a standard multi-class task for which one can readily train a classifier using standard techniques. However, training data for such scenarios is scarce and it would be costly and time consuming to collect for each question and topic. Our solution is to break each topic into granular components that can be identified using large language models (LLMs) with a straightforward generic tuning.

We map each topic to a list of underlying questions and define that if the sentence contains an answer to one of those underlying questions, then it covers that topic. For the topic “Experience” we might choose underlying questions such as:

  • Where did you work?
  • What did you study?

While for the topic “Interests” we might choose underlying questions such as:

  • What are you interested in?
  • What do you enjoy doing?

These underlying questions are designed through an iterative manual process. Importantly, since these questions are sufficiently granular, current language models (see details below) can capture their semantics. This allows us to offer a zero-shot setting for the NLA topicality task: once trained (more on the model below), it is easy to add new questions and new topics, or adapt existing topics by modifying their underlying content expectation without the need to collect topic specific data. See below the model’s predictions for the sentence “I’ve worked in retail for 3 years” for the two topics described above:

A diagram of how the model uses underlying questions to predict the topic most likely to be covered by the user’s answer.

Since an underlying question for the topic “Experience” was matched, the sentence would be classified as “Experience”.

Application: Helping Job Seekers Prepare for Interviews

Interview Warmup is a new tool developed in collaboration with job seekers to help them prepare for interviews in fast-growing fields of employment such as IT Support and UX Design. It allows job seekers to practice answering questions selected by industry experts and to become more confident and comfortable with interviewing. As we worked with job seekers to understand their challenges in preparing for interviews and how an interview practice tool could be most useful, it inspired our research and the application of topicality NLA.

We build the topicality NLA model (once for all questions and topics) as follows: we train an encoder-only T5 model (EncT5 architecture) with 350 million parameters on Question-Answers data to predict the compatibility of an <underlying question, answer> pair. We rely on data from SQuAD 2.0 which was processed to produce <question, answer, label> triplets.

In the Interview Warmup tool, users can switch between talking points to see which ones were detected in their answer.

The tool does not grade or judge answers. Instead it enables users to practice and identify ways to improve on their own. After a user replies to an interview question, their answer is parsed sentence-by-sentence with the Topicality NLA model. They can then switch between different talking points to see which ones were detected in their answer. We know that there are many potential pitfalls in signaling to a user that their response is “good”, especially as we only detect a limited set of topics. Instead, we keep the control in the user’s hands and only use ML to help users make their own discoveries about how to improve.

So far, the tool has had great results helping job seekers around the world, including in the US, and we have recently expanded it to Africa. We plan to continue working with job seekers to iterate and make the tool even more helpful to the millions of people searching for new jobs.

A short film showing how Interview Warmup and its NLA capabilities were developed in collaboration with job seekers.


Natural Language Assessment (NLA) is a technologically challenging and interesting research area. It paves the way for new conversational applications that promote learning by enabling the nuanced assessment and analysis of answers from multiple perspectives. Working together with communities, from job seekers and businesses to classroom teachers and students, we can identify situations where NLA has the potential to help people learn, engage, and develop skills across an array of subjects, and we can build applications in a responsible way that empower users to assess their own abilities and discover ways to improve.


This work is made possible through a collaboration spanning several teams across Google. We’d like to acknowledge contributions from Google Research Israel, Google Creative Lab, and Grow with Google teams among others.

Source: Google AI Blog

Digitizing Smell: Using Molecular Maps to Understand Odor

Did you ever try to measure a smell? …Until you can measure their likenesses and differences you can have no science of odor. If you are ambitious to found a new science, measure a smell.
— Alexander Graham Bell, 1914.

How can we measure a smell? Smells are produced by molecules that waft through the air, enter our noses, and bind to sensory receptors. Potentially billions of molecules can produce a smell, so figuring out which ones produce which smells is difficult to catalog or predict. Sensory maps can help us solve this problem. Color vision has the most familiar examples of these maps, from the color wheel we each learn in primary school to more sophisticated variants used to perform color correction in video production. While these maps have existed for centuries, useful maps for smell have been missing, because smell is a harder problem to crack: molecules vary in many more ways than photons do; data collection requires physical proximity between the smeller and smell (we don’t have good smell “cameras” and smell “monitors”); and the human eye only has three sensory receptors for color while the human nose has > 300 for odor. As a result, previous efforts to produce odor maps have failed to gain traction.

In 2019, we developed a graph neural network (GNN) model that began to explore thousands of examples of distinct molecules paired with the smell labels that they evoke, e.g., “beefy”, “floral”, or “minty”, to learn the relationship between a molecule’s structure and the probability that such a molecule would have each smell label. The embedding space of this model contains a representation of each molecule as a fixed-length vector describing that molecule in terms of its odor, much as the RGB value of a visual stimulus describes its color.

Left: An example of a color map (CIE 1931) in which coordinates can be directly translated into values for hue and saturation. Similar colors lie near each other, and specific wavelengths of light (and combinations thereof) can be identified with positions on the map. Right: Odors in the Principal Odor Map operate similarly. Individual molecules correspond to points (grey), and the locations of these points reflect predictions of their odor character.

Today we introduce the “Principal Odor Map” (POM), which identifies the vector representation of each odorous molecule in the model’s embedding space as a single point in a high-dimensional space. The POM has the properties of a sensory map: first, pairs of perceptually similar odors correspond to two nearby points in the POM (by analogy, red is nearer to orange than to green on the color wheel). Second, the POM enables us to predict and discover new odors and the molecules that produce them. In a series of papers, we demonstrate that the map can be used to prospectively predict the odor properties of molecules, understand these properties in terms of fundamental biology, and tackle pressing global health problems. We discuss each of these promising applications of the POM and how we test them below.

Test 1: Challenging the Model with Molecules Never Smelled Before
First, we asked if the underlying model could correctly predict the odors of new molecules that no one had ever smelled before and that were very different from molecules used during model development. This is an important test — many models perform well on data that looks similar to what the model has seen before, but break down when tested on novel cases.

To test this, we collected the largest ever dataset of odor descriptions for novel molecules. Our partners at the Monell Center trained panelists to rate the smell of each of 400 molecules using 55 distinct labels (e.g., “minty”) that were selected to cover the space of possible smells while being neither redundant nor too sparse. Unsurprisingly, we found that different people had different characterizations of the same molecule. This is why sensory research typically uses panels of dozens or hundreds of people and highlights why smell is a hard problem to solve. Rather than see if the model could match any one person, we asked how close it was to the consensus: the average across all of the panelists. We found that the predictions of the model were closer to the consensus than the average panelist was. In other words, the model demonstrated an exceptional ability to predict odor from a molecule’s structure.

Predictions made by two models, our GNN model (orange) and a baseline chemoinformatic random forest (RF) model (blue), compared with the mean ratings given by trained panelists (green) for the molecule 2,3-dihydrobenzofuran-5-carboxaldehyde. Each bar corresponds to one odor character label (with only the top 17 of 55 shown for clarity). The top five are indicated in color; our model correctly identifies four of the top five, with high confidence, vs. only three of five, with low confidence, for the RF model. The correlation (R) to the full set of 55 labels is also higher in our model.
Unlike alternative benchmark models (RF and nearest-neighbor models trained on various sets of chemoinformatic features), our GNN model outperforms the median human panelist at predicting the panel mean rating. In other words, our GNN model better reflects the panel consensus than the typical panelist.

The POM also exhibited state-of-the-art performance on alternative human olfaction tasks like detecting the strength of a smell or the similarity of different smells. Thus, with the POM, it should be possible to predict the odor qualities of any of billions of as-yet-unknown odorous molecules, with broad applications to flavor and fragrance.

Test 2: Linking Odor Quality Back to Fundamental Biology
Because the Principal Odor Map was useful in predicting human odor perception, we asked whether it could also predict odor perception in animals, and the brain activity that underlies it. We found that the map could successfully predict the activity of sensory receptors, neurons, and behavior in most animals that olfactory neuroscientists have studied, including mice and insects.

What common feature of the natural world makes this map applicable to species separated by hundreds of millions of years of evolution? We realized that the common purpose of the ability to smell might be to detect and discriminate between metabolic states, i.e., to sense when something is ripe vs. rotten, nutritious vs. inert, or healthy vs. sick. We gathered data about metabolic reactions in dozens of species across the kingdoms of life and found that the map corresponds closely to metabolism itself. When two molecules are far apart in odor, according to the map, a long series of metabolic reactions is required to convert one to the other; by contrast, similarly smelling molecules are separated by just one or a few reactions. Even long reaction pathways containing many steps trace smooth paths through the map. And molecules that co-occur in the same natural substances (e.g., an orange) are often very tightly clustered on the map. The POM shows that olfaction is linked to our natural world through the structure of metabolism and, perhaps surprisingly, captures fundamental principles of biology.

Left: We aggregated metabolic reactions found in 17 species across 4 kingdoms to construct a metabolic graph. In this illustration, each circle is a distinct metabolite molecule and an arrow indicates that there is a metabolic reaction that converts one molecule to another. Some metabolites have an odor (color) and others do not (gray), and the metabolic distance between two odorous metabolites is the minimum number of reactions necessary to convert one into the other. In the path shown in bold, the distance is 3. Right: Metabolic distance was highly correlated with distance in the POM, an estimate of perceived odor dissimilarity.

Test 3: Extending the Model to Tackle a Global Health Challenge
A map of odor that is tightly connected to perception and biology across the animal kingdom opens new doors. Mosquitos and other insect pests are drawn to humans in part by their odor perception. Since the POM can be used to predict animal olfaction generally, we retrained it to tackle one of humanity’s biggest problems, the scourge of diseases transmitted by mosquitoes and ticks, which kill hundreds of thousands of people each year.

For this purpose, we improved our original model with two new sources of data: (1) a long-forgotten set of experiments conducted by the USDA on human volunteers beginning 80 years ago and recently made discoverable by Google Books, which we subsequently made machine-readable; and (2) a new dataset collected by our partners at TropIQ, using their high-throughput laboratory mosquito assay. Both datasets measure how well a given molecule keeps mosquitos away. Together, the resulting model can predict the mosquito repellency of nearly any molecule, enabling a virtual screen over huge swaths of molecular space. We validated this screen experimentally using entirely new molecules and found over a dozen of them with repellency at least as high as DEET, the active ingredient in most insect repellents. Less expensive, longer lasting, and safer repellents can reduce the worldwide incidence of diseases like malaria, potentially saving countless lives.

We digitized USDA mosquito repellency data for thousands of molecules previously scanned by Google Books, and used it to refine the learned representation (the map) at the heart of the model. We added additional layers, specifically to predict repellency in a mosquito feeder assay, and iteratively trained the model to improve assay predictions while running computational screens for candidate repellents.
Many molecules showing mosquito repellency in the laboratory assay also showed repellency when applied to humans. Several showed repellency greater than the most common repellents used today (DEET and picaridin).

The Road Ahead
We discovered that our modeling approach to smell prediction could be used to draw a Principal Odor Map for tackling odor-related problems more generally. This map was the key to measuring smell: it answered a range of questions about novel smells and the molecules that produce them, it connected smells back to their origins in evolution and the natural world, and it is helping us tackle important human-health challenges that affect millions of people. Going forward, we hope that this approach can be used to find new solutions to problems in food and fragrance formulation, environmental quality monitoring, and the detection of human and animal diseases.

This work was performed by the ML olfaction research team, including Benjamin Sanchez-Lengeling, Brian K. Lee, Jennifer N. Wei, Wesley W. Qian, and Jake Yasonik (the latter two were partly supported by the Google Student Researcher program) and our external partners including Emily Mayhew and Joel D. Mainland from the Monell Center, and Koen Dechering and Marnix Vlot from TropIQ. The Google Books team brought the USDA dataset online. Richard C. Gerkin was supported by the Google Visiting Faculty Researcher program and is also an Associate Research Professor at Arizona State University.

Source: Google AI Blog