Robotic Arm Manipulation with Inverse Reinforcement Learning & TD-MPC

TL;DR

This paper introduces a gradient-based inverse reinforcement learning framework that learns cost functions from visual demonstrations and applies TD-MPC for robotic manipulation, addressing scalability and dynamic challenges.

Contribution

It presents a novel IRL method that learns from visual data and integrates with TD-MPC for effective robotic manipulation in high-dimensional spaces.

Findings

Successful demonstration on hardware object manipulation tasks

Effective learning of cost functions from visual demonstrations

Integration of IRL with TD-MPC improves control performance

Abstract

One unresolved issue is how to scale model-based inverse reinforcement learning (IRL) to actual robotic manipulation tasks with unpredictable dynamics. The ability to learn from both visual and proprioceptive examples, creating algorithms that scale to high-dimensional state-spaces, and mastering strong dynamics models are the main obstacles. In this work, we provide a gradient-based inverse reinforcement learning framework that learns cost functions purely from visual human demonstrations. The shown behavior and the trajectory is then optimized using TD visual model predictive control(MPC) and the learned cost functions. We test our system using fundamental object manipulation tasks on hardware.