Collaborative Assembly Policy Learning of a Sightless Robot

2025-11-05

AI Summary

Blind robotic execution of frame-insertion tasks under vision-free conditions suffers from inaccurate human force/torque measurements in conventional admittance control—leading to erroneous intent estimation—and faces challenges in applying reinforcement learning (RL) due to stringent safety constraints and sparse reward signals. Method: This paper proposes an admittance-guided RL framework, leveraging a handcrafted admittance controller as a prior policy to constrain the RL policy search space, thereby ensuring safety while improving exploration efficiency. The method integrates real-time force/torque sensory feedback and employs joint training across simulation and physical platforms. Results: Experiments demonstrate that, compared to pure admittance control, the proposed approach increases task success rate by 32%, reduces average completion time by 27%, and lowers peak interaction force/torque by 41%. It significantly enhances robotic proactivity and alleviates human operator workload.

Technology Category

• Intelligent Robots: Manipulation
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

This paper explores a physical human-robot collaboration (pHRC) task involving the joint insertion of a board into a frame by a sightless robot and a human operator. While admittance control is commonly used in pHRC tasks, it can be challenging to measure the force/torque applied by the human for accurate human intent estimation, limiting the robot's ability to assist in the collaborative task. Other methods that attempt to solve pHRC tasks using reinforcement learning (RL) are also unsuitable for the board-insertion task due to its safety constraints and sparse rewards. Therefore, we propose a novel RL approach that utilizes a human-designed admittance controller to facilitate more active robot behavior and reduce human effort. Through simulation and real-world experiments, we demonstrate that our approach outperforms admittance control in terms of success rate and task completion time. Additionally, we observed a significant reduction in measured force/torque when using our proposed approach compared to admittance control. The video of the experiments is available at https://youtu.be/va07Gw6YIog.

Problem

• Developing collaborative assembly policy for sightless robot-human board insertion task
• Overcoming admittance control limitations in human intent estimation
• Addressing safety constraints and sparse rewards in reinforcement learning

Innovation

• Reinforcement learning with admittance controller guidance
• Active robot behavior reduces human physical effort
• Simulation and real-world validation show performance improvement