Failure Forecasting Boosts Robustness of Sim2Real Rhythmic Insertion Policies

Yuhan Liu, Xinyu Zhang, Haonan Chang, Abdeslam Boularias
Rutgers University
IROS 2025
Code arXiv

Our proposed failure forecasting and recovery algorithm significantly increases the robustness of Rhythmic Insertion policies like screwing a nut into a bolt with a wrench.

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.

System Pipeline Illustration

Robot executing wrench-based nut insertion: employing a trained neural network to forecast potential failures, the system autonomously corrects its trajectory and retries until the nut is fully inserted, exemplifying enhanced robustness of rhythmic insertion policies.

RIT Illustration

Proposed Method

The proposed method consists of three components: (1) a RL-based single-time insertion policy, (2) a failure forecasting model, and (3) a recovery mechanism.

RIT System Pipeline
System Overview
RIT Failure Forecasting Models
Failure Forecasting Models

Video Presentation

BibTeX

@misc{liu2025failureforecastingboostsrobustness,
        title={Failure Forecasting Boosts Robustness of Sim2Real Rhythmic Insertion Policies}, 
        author={Yuhan Liu and Xinyu Zhang and Haonan Chang and Abdeslam Boularias},
        year={2025},
        eprint={2507.06519},
        archivePrefix={arXiv},
        primaryClass={cs.RO},
        url={https://arxiv.org/abs/2507.06519}, 
  }