Papers
2
Total Citations
20
H-Index
2
About
Parv Kapoor is a researcher at the frontier of autonomous systems, specializing in safe reinforcement learning (RL), formal methods, and human-robot teaming. His work addresses a critical gap: how to guarantee that learning-based control policies obey complex, time-sensitive safety rules. In his highly cited 2020 paper, Kapoor pioneered model-based RL from Signal Temporal Logic (STL) specifications, offering a principled alternative to fragile reward functions. This approach prevents the common RL pitfall of reward hacking, ensuring that robots learn behaviors that are both optimal and verifiably safe—a contribution that has earned 14 citations and growing interest from the robotics community. Kapoor also tackles the pressing challenge of integrating autonomous unmanned aircraft with manned traffic in shared airspace. His 2022 work proposes a framework for close-proximity, safe, and seamless operation, enabling manned-unmanned vehicle teaming where each agent learns from the other. By combining formal logic with learning, Kapoor is building a foundation for trustworthy autonomy in high-stakes environments, from aerial logistics to urban mobility.
Research Focus
Key Achievements
Top Papers
- 1Model-based Reinforcement Learning from Signal Temporal Logic Specifications14 citations · 2020
- 2