Vaisakh Shaj
Papers
2
Total Citations
14
H-Index
2
About
Vaisakh Shaj is a researcher at the forefront of merging machine learning with model-based robot control, specializing in hybrid inverse dynamics and state estimation. His work addresses a critical challenge in robotics: achieving both precise tracking and compliant interaction by combining analytical rigid-body models with data-driven learning for hard-to-model effects like stick-slip friction. His most cited paper, "End-to-End Learning of Hybrid Inverse Dynamics Models for Precise and Compliant Impedance Control" (2022, 11 citations), demonstrates how neural networks can seamlessly augment classical dynamics to improve impedance control performance. Shaj further advanced the field with "Action-Conditional Recurrent Kalman Networks For Forward and Inverse Dynamics Learning" (2020, 3 citations), introducing a novel architecture that integrates recurrent neural networks with Kalman filtering to estimate dynamics in complex systems—such as hydraulically actuated robots or those with artificial muscles—where analytic models are unavailable. By enabling robots to learn accurate dynamics from data while retaining the structure of known physics, Shaj’s contributions are paving the way for more adaptive, compliant, and reliable robotic systems in real-world applications.
Research Focus
Key Achievements
Top Papers
- 1
- 2