Kavya  Anbarasu

Kavya Anbarasu

Scholar Title

Undergraduate Research and Innovation Scholar

Research Title

Training Policies for Robotic Hand Exoskeletons

Cohort

2024–2025

Department

Electrical Engineering and Computer Science

Research Areas
  • Robotics
Supervisor

Pulkit Agrawal

Abstract

This project explores the use of transfer learning to develop advanced control policies for robotic manipulation based on human-exoskeleton interactions. By capturing rich motion and tactile data from human users, the project trains neural networks that can replicate the dexterity of human hands. These trained policies are then transferred to robotic systems, allowing them to perform complex tasks in dynamic, unstructured environments. The use of transfer learning enables the adaptation of these policies across different robotic platforms, improving generalization and efficiency. By focusing on policy transfer from human demonstrations to robots, the project aims to overcome limitations of teleoperation and synthetic simulations, providing scalable solutions for training robots to perform a wide variety of tasks with precision and adaptability.

Quote

I’m really, really excited by the future of robotics. I think the project I am working on will help enable large scale data collection and policy transfer on robots and I think that is the next step in the field (and towards AGI :0). Also, I am super excited to be able to apply what I learnt in class and in textbooks about reinforcement learning in the lab and get more hands on experience.

Back to Scholars