Design for Safety
Methods for safety-aware data collection, policy training, verification, and evaluation that help learning-enabled autonomous systems avoid unsafe behavior before deployment.
Safe and Intelligent Autonomy Lab
Welcome to the Safe and Intelligent Autonomy (SIA) Lab! SIA Lab is part of the Department of Aeronautics and Astronautics at Stanford. Our lab develops principled methods for ensuring the safety of learning-enabled autonomous systems.
Modern robots increasingly rely on machine learning, foundation models, and data-driven decision-making to perceive, plan, and act in complex environments. These capabilities create tremendous opportunities, but also introduce new safety challenges: failures can arise from imperfect models, distribution shift, brittle policies, untested scenarios, and unexpected interactions with the real world.
Our central goal is to build the safety lifecycle for learning-enabled autonomy. We develop theory, algorithms, and computational tools that help autonomous systems discover failures before deployment, intervene safely during deployment, and improve from failures and near-misses after deployment.
The technical backbone of our work is principled safety: reachability analysis, control theory, planning, verification, and learning-based safety representations. We use these foundations to design safer policies, evaluate their limitations, deploy them with runtime safeguards, and continually adapt them to new safety risks.
Research Themes
Intervene Safely
Runtime monitors, safety filters, reachability-guided planning, and recovery mechanisms that help autonomous systems respond safely to uncertainty and emerging risks.
Improve Over Time
Failure analysis, near-miss mining, and safety feedback loops that use deployment experience to improve policies, monitors, simulators, and safety models.
Safety Foundations
Foundations in reachability, control theory, planning, verification, and learning that provide rigorous tools for reasoning about safety in autonomous systems.
News
October 2026
Three new preprints on world model safety, reasoning steering and context-conditioned HJ reachability.
September 2026
Congrats Hao on winning the IEEE IROS 2026 Workshop on Deformable Objects Manipulation (DOM-R3) best paper award!
September 2026
SIA Lab has 3 new PhD Students: John Cao, Thomas MacLean, Bryan Boateng, and 1 new visiting student: Noé Macé. Welcome!
September 2026
Congrats Zeyuan on his paper acceptance at CoRL 2026!
August 2026
We have been selected to lead a new NASA University Leadership Initiative (ULI) Center on 𝐒𝐚𝐟𝐞𝐭𝐲 𝐀𝐜𝐫𝐨𝐬𝐬 𝐭𝐡𝐞 𝐋𝐢𝐟𝐞𝐜𝐲𝐜𝐥𝐞 𝐨𝐟 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠-𝐄𝐧𝐚𝐛𝐥𝐞𝐝 𝐀𝐯𝐢𝐨𝐧𝐢𝐜𝐬 𝐒𝐲𝐬𝐭𝐞𝐦𝐬 (𝐋𝐄𝐀𝐒)!
June 2026
Congrats to Somil for winning the AIAA Excellence in Teaching Award!
June 2026
The SIA lab has a new rotating PhD student, Thomas MacLean, a visiting postdoc, Ajay Suresha Sathya, and two new visiting students Yunus Yazoglu from ETH and Akira Hatakeyama from JAXA. Welcome!
June 2026
Congrats to Umut, Aryaman, and Zeyuan for starting internships with Toyota Research Institute, Honda Research Institute, and Amazon Robotics.
June 2026
Four papers from SIA Lab were presented at ICRA 2026. Congratulations to Albert, Javier, Kaustav, Ryan, and collaborators.
May 2026
Congratulations to Dr. Kaustav Chakraborty for becoming the second member to graduate from our lab, and to Dr. Javier Borquez on joining Universidad de Santiago de Chile as faculty.
May 2026
Hao, Javier, Jason, Zeyuan, and collaborators will present a tutorial on reachability and CBF at CDC 2026.
May 2026
Congratulations to Sathwik and Aryaman for winning the Qualcomm Innovation Fellowship.
May 2026
Congratulations to Dr. Somil Bansal for receiving the IEEE Early Career Award in Robotics and Automation from the Robotics and Automation Society.
April 2026
Colton received the inaugural Stanford Robotics Fellowship to develop reliable monitoring and intervention schemes for VLAs.
April 2026
Congratulations Hao and Colton for the paper acceptance to RSS 2026.
April 2026
We started a three-year project with Toyota Research Institute focusing on safety in high-speed driving.
March 2026
The SIA lab has a new visiting student, Ali Fuat Sahin from EPFL, and an MS student member Joseph Lee from Stanford. Welcome!
March 2026
Congratulations Le and Umut for your paper acceptance to IEEE Robotics and Automation Letters.
January 2026
The SIA lab has a new rotating PhD student, Jearyoung Lee. Welcome!
December 2025
Congratulations Albert for your paper acceptance to L4DC 2026.
September 2025
New paper on detecting and steering misalignment in LLMs via latent reachability.