Papers
3
Total Citations
10
H-Index
2
About
Amit Parag is a researcher at the forefront of reinforcement learning and robotic manipulation, specializing in the intersection of differentiable simulation, trajectory optimization, and tactile sensing. His work addresses fundamental challenges in enabling robots to learn complex control tasks with greater efficiency and reliability. Parag’s major contributions include pioneering a hybrid approach that integrates trajectory optimization with Sobolev descent, achieving superlinear convergence properties in reinforcement learning—a significant step toward reducing the massive data requirements typical of deep RL methods. This work has garnered 6 citations and is foundational for developing sample-efficient learning algorithms. More recently, he has advanced robotic grasping by applying Video Vision Transformers to detect incipient slip using GelSight tactile sensors, a breakthrough that enhances real-time dexterous manipulation and has already attracted 3 citations. His 2025 paper on optimizing complex control systems with differentiable simulators further cements his impact, offering a novel framework that leverages gradient information for more precise trajectory planning. With a growing citation record and a focus on practical, data-efficient solutions, Parag is shaping the future of autonomous robotic systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3