Failure Forecasting Boosts Robustness of Sim2Real Rhythmic Insertion Policies

2025-07-08

AI Summary

Robotic execution of high-precision rhythmic insertion tasks—e.g., nut threading onto bolts—is hindered by friction, rotational dynamics, and sim-to-real domain gaps, leading to degraded repeatability and long-term stability. Method: We propose a Sim2Real framework integrating reinforcement learning (RL) with lightweight failure prediction. Crucially, the wrench’s 6D pose is represented in the target object’s (nut’s) coordinate frame to improve policy transferability; real-time 6D pose tracking enables precise alignment, insertion, and rotation; and a compact neural failure predictor dynamically triggers a lift-and-retry recovery mechanism. Contribution/Results: The method achieves >95% single-trial success rates in both simulation and physical experiments. It maintains consistent performance over >100 consecutive trials, demonstrating superior robustness against complex contact disturbances and significantly enhancing long-horizon task reliability.

Technology Category

• Intelligent Robots: Manipulation
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

This paper addresses the challenges of Rhythmic Insertion Tasks (RIT), where a robot must repeatedly perform high-precision insertions, such as screwing a nut into a bolt with a wrench. The inherent difficulty of RIT lies in achieving millimeter-level accuracy and maintaining consistent performance over multiple repetitions, particularly when factors like nut rotation and friction introduce additional complexity. We propose a sim-to-real framework that integrates a reinforcement learning-based insertion policy with a failure forecasting module. By representing the wrench's pose in the nut's coordinate frame rather than the robot's frame, our approach significantly enhances sim-to-real transferability. The insertion policy, trained in simulation, leverages real-time 6D pose tracking to execute precise alignment, insertion, and rotation maneuvers. Simultaneously, a neural network predicts potential execution failures, triggering a simple recovery mechanism that lifts the wrench and retries the insertion. Extensive experiments in both simulated and real-world environments demonstrate that our method not only achieves a high one-time success rate but also robustly maintains performance over long-horizon repetitive tasks.

Problem

• Achieving millimeter-level accuracy in rhythmic insertion tasks
• Maintaining consistent performance despite rotation and friction
• Enhancing sim-to-real transferability for robotic insertion policies

Innovation

• Sim-to-real framework with failure forecasting
• 6D pose tracking for precise alignment
• Neural network predicts execution failures