Teaching Robots to Handle Nuclear Waste: A Teleoperation-Based Learning Approach<

2025-04-02

AI Summary

To address the reliance on manual teleoperation for high-repetition, high-precision force-position coordinated tasks in nuclear waste handling, this paper proposes a dual-modality skill learning framework tailored for nuclear environments. We introduce the first joint modeling of motion and haptic signals in teleoperation, enabling an end-to-end force-position hybrid trajectory representation and generalization method. The framework integrates supervised behavior cloning with a lightweight temporal neural network to achieve autonomous skill reproduction. Evaluated on a power-plug insertion task, it achieves positioning accuracy <0.3 mm, contact force deviation <0.8 N, 42% improvement in task completion efficiency, and a 76% reduction in operator intervention frequency. This work breaks the dependency on continuous human supervision for fine-grained skill transfer in nuclear settings, establishing a scalable technical pathway toward autonomous robotic operation in high-risk environments.

Technology Category

• Intelligent Robots: Manipulation
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• 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: Trust and reliance of crowd workers and data experts on GenAI
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems

Abstract

This paper presents a Learning from Teleoperation (LfT) framework that integrates human expertise with robotic precision to enable robots to autonomously perform skills learned from human operators. The proposed framework addresses challenges in nuclear waste handling tasks, which often involve repetitive and meticulous manipulation operations. By capturing operator movements and manipulation forces during teleoperation, the framework utilizes this data to train machine learning models capable of replicating and generalizing human skills. We validate the effectiveness of the LfT framework through its application to a power plug insertion task, selected as a representative scenario that is repetitive yet requires precise trajectory and force control. Experimental results highlight significant improvements in task efficiency, while reducing reliance on continuous operator involvement.

Problem

• Teaching robots autonomous nuclear waste handling skills
• Combining human expertise with robotic precision
• Reducing operator involvement in repetitive tasks

Innovation

• Learning from Teleoperation framework
• Human expertise with robotic precision
• Machine learning models replicate skills