Abhinav Bhatia
Papers
2
Total Citations
12
H-Index
2
About
Abhinav Bhatia is a researcher at the forefront of integrating metareasoning with deep reinforcement learning to enhance decision-making in complex, real-time systems. His primary research areas lie in anytime planning, Markov decision processes (MDPs), and state abstraction, with a specific focus on how autonomous agents can intelligently manage their own computational resources. Bhatia’s major contribution is expanding the scope of metareasoning beyond simply deciding *when* to act; he has pioneered methods for dynamically tuning the hyperparameters of anytime planning algorithms at runtime. His 2022 paper on this topic, which has garnered 9 citations, demonstrates how deep reinforcement learning can optimize these parameters to improve plan quality under time pressure. In a complementary 2022 work (3 citations), he applied a similar metareasoning framework to the challenge of selecting partial state abstractions for MDPs, enabling robots to intelligently reduce problem complexity without sacrificing solution quality. By teaching agents to reason about their own reasoning processes, Bhatia’s work is paving the way for more adaptive and efficient autonomous systems in robotics and AI.
Research Focus
Key Achievements
Top Papers
- 1
- 2