A Learning Approach to Robot-Agnostic Force-Guided High Precision Assembly

This Autodesk Research paper provides the following impact to the field of robotics:

• It provides an RL-based robotic assembly system that is environment-agnostic and requires no vision system, motion capture or other tracking system, which is an important factor to practice AI-enhanced robotic assembly in uncertain environments, like construction sites.
• By reducing the number of sensing sources needed in the task to the minimum, we reduce the number of sources that contribute to the sim-to-real gap, and therefore, improve learning transfer from simulation to reality.
• This research can potentially be extended to other robotic tasks that are partially observable. For example, one of the issues in Tesla’s fully robotic assembly line is the robots are vision-impaired due to countless unexpected occlusions in assembly.
• To our knowledge, this will the first successful application of reinforcement learning on robotic assembly that only uses torque/force observations.

Abstract

A Learning Approach to Robot-Agnostic Force-Guided High Precision Assembly

Jieliang Luo, Hui Li

IEEE/RSJ International Conference on Intelligent Robots and System 2021

In this work we propose a learning approach to high-precision robotic assembly problems in the continuous action domain. Unlike many learning-based approaches that heavily rely on vision or spatial tracking, our approach takes force/torque as the only observation. Each learned policy from our approach is robot-agnostic, which can be applied to different robotic arms. These two features can greatly reduce complexity and cost to perform robotic assembly in the real world, especially in unstructured settings such as in architectural construction. To achieve it, we have developed a new distributed RL agent, named Recurrent Distributed DDPG (RD2), which extends Ape-X DDPG with recurrency and makes two structural improvements on prioritized experience replay. Our results show that RD2 is able to solve two fundamental high-precision assembly tasks, lap-joint and peg-in-hole, and outperforms two state-of-the-art algorithms, Ape-X DDPG and PPO with LSTM. We have successfully evaluated our robot-agnostic policies on three robotic arms, Kuka KR60, Franka Panda, and UR10, in simulation.