The Safe and Intelligent Autonomy Lab is part of the Department of Aeronautics and Astronautics at Stanford.
We seek to design autonomous systems and robots that operate with guaranteed safety and performance in new and uncertain environments. The current focus of the lab is to ensure safety when these systems leverage machine learning and AI for decision-making and control.
To achieve this, our research emphasizes developing fundamental frameworks and computational tools that integrate safety into every stage of the learning process. This includes methods to provide safety assurances during the design/training phase (e.g., via safe learning and correct-by-construction policies); continuous safety monitoring during operation (e.g., through uncertainty quantification and anomaly detection); and dynamically adapting to new safety risks during the system’s lifetime (e.g., via developing new safety and reachability methods for nonlinear and high-dimensional system). Algorithmically, our research blends classical control theory and modern machine learning methods towards this goal, with applications across robotics, surveillance, air traffic management, space exploration, and autonomous driving.
Please see this talk for a nice summary of the latest research directions in our group.
We are looking for undergraduate and graduate students to join our research group. If you are interested in robotics, control, and/or machine learning, consider joining us!
News
[December 2024] Congratulations Hao and Adityaya, for your paper acceptance to IEEE Control Systems Letters (L-CSS).
[December 2024] New paper: "One Filter to Deploy Them All: Robust Safety for Quadrupedal Navigation in Unknown Environments" is out on arXiv.
[October 2024] New paper on Enhancing Safety and Robustness of Vision-Based Controllers via Reachability Analysis is out on arXiv.
[September 2024] Six new pre-prints have been posted on arXiv.
[July 2024] Two papers accepted at CDC 2024! Congratulations.
[August 2024] Congratulations Javier, for your paper acceptance to IEEE Transactions on Robotics.
[June 2024] Congratulations Hao, for your paper acceptance to IEEE Control Systems Letters.
[June 2024] Congrats to Javier for the paper on Hamilton-Jacobi reachability analysis for hybrid systems with controlled and forced transitions acceptance to RSS 2024.
[April 2024] Congratulations Albert, for your paper acceptance to L4DC 2024
[April 2024] New paper on Safety and Liveness Filtering Using Hamilton-Jacobi Reachability Analysis.
[April 2024] New paper on Providing Safety Assurances for Systems with Unknown Dynamics is out on arXiv.
[April 2024] New paper, SAFE-GIL: SAFEty Guided Imitation Learning, is out on arXiv.
[April 2024] New paper on Imposing Exact Safety Specifications in Neural Reachable Tubes is out on arXiv.
[December 2023] New paper on verification of neural reachable tubes via scenario optimization and conformal prediction is out on arXiv.
[September 2023] New paper on detecting and mitigating system-level anomalies of vision-based controllers is out on arXiv.
[September 2023] New paper on Hamilton-Jacobi reachability analysis for hybrid systems with controlled and forced transitions is out on arXiv.
[August 2023] Three undergraduate students have joined our lab through the CURVE program. Welcome Kyle, Saim, and Sascha!