Kishor Jothimurugan
Papers
1
Total Citations
21
H-Index
1
About
Kishor Jothimurugan is a researcher advancing the frontier of reinforcement learning through formal methods and compositional task specification. His work addresses a critical bottleneck in deploying RL for robotics: the difficulty of designing reward functions that correctly encode complex, multi-objective tasks with safety constraints. His most-cited paper, "A Composable Specification Language for Reinforcement Learning Tasks" (2020, 21 citations), introduces a structured language that allows users to break down intricate robot control problems into modular, verifiable components. This approach not only simplifies task design but also enhances interpretability and safety, enabling more reliable learning of policies. By bridging formal verification and reinforcement learning, Jothimurugan’s contributions are pivotal for creating autonomous systems that can handle real-world complexity. His work is essential reading for researchers seeking principled methods to specify and learn high-stakes robotic behaviors, demonstrating how compositionality can transform RL from a black-box optimization into a transparent, engineerable discipline.
Research Focus
Key Achievements
Top Papers
- 1A Composable Specification Language for Reinforcement Learning Tasks21 citations · 2020