Jakob Thumm
Papers
3
Total Citations
65
H-Index
3
About
Jakob Thumm is a leading researcher at the intersection of safe artificial intelligence and human-robot collaboration, whose work is fundamentally reshaping how robots can operate safely alongside people. His primary research focuses on developing provably safe deep reinforcement learning (RL) for robotic manipulation, with a particular emphasis on handling highly dynamic obstacles like humans. Thumm’s landmark paper, “Provably Safe Deep Reinforcement Learning for Robotic Manipulation in Human Environments” (2022, 37 citations), addresses the critical gap in formal safety assurances for RL-based manipulator control, offering a method that guarantees safety even in unpredictable human environments. He further advanced the field with “SaRA: A Tool for Safe Human-Robot Coexistence and Collaboration through Reachability Analysis” (2022, 19 citations), which provides formal safety guarantees for human-robot interaction beyond traditional caging methods. Most recently, Thumm introduced the “Human-Robot Gym” (2024, 9 citations), a benchmark designed to standardize the evaluation of RL approaches in human-robot collaboration under safety constraints. His work is pivotal for enabling the widespread adoption of collaborative robots in manufacturing, healthcare, and domestic settings, where safe, adaptive, and intelligent robot behavior is essential.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3