Welcome to the Safe Robotics Lab @ GT!
We build full-stack robotic systems that are not only safe, but also SAFE: Smart, Agile, Flexible, and Engaging. Our long-term goal is to enable widespread, safe robot deployment to understand how, when, and where society wants and needs robots.
The SRL explores, develops, and implements methods in every part of a robot’s autonomy stack: hardware, perception, planning, and control. We seek to decouple the safety and performance tradeoff, making robots that can not just accomplish difficult tasks but also guarantee that they meet constraints while doing so. To accomplish this goal we use novel designs, formal methods, and learning techniques to ensure that uncertain autonomous systems can safely achieve tasks to benefit humanity.
Robotics should be by everyone, for everyone.
The Safe Robotics Lab strives to not just build better robots,
but also better roboticists.
Latest News
- 2025/09/27: We have a new paper on using LLMs to coordinate warehouse robots.
- 2025/09/17: We have a new paper on using LLMs to predict aircraft intentions.
- 2025/08/01: Two papers accepted to CoRL: JM2D (A unified diffusion framework for model-free planning and model-based optimization) and SAIL (Faster-than-demonstrator Imitation Learning Policy)!
- 2025/07/15: Our paper on training neural network with hybrid zonotopes has been accepted to IEEE Conference on Decision and Control (CDC)!
- 2025/06/30: Our new paper on provably safe manipulator motion planning, in collaboration with Prof. Ram Vasudevan, is available in the IEEE Transactions on Robotics!
See all past updates here.