Self-supervised Physics-Informed Manipulation of Deformable Linear Objects with Non-negligible Dynamics

2026-02-03

AI Summary

This work addresses the challenge of robust manipulation of dynamic deformable linear objects—such as ropes—under strong dynamical effects and state noise. The authors propose SPiD, a framework that integrates a lightweight differentiable mass-spring physical model with self-supervised learning to enable end-to-end training of a neural controller. A novel self-supervised variant of DAgger is introduced to detect distribution shifts and facilitate offline self-correction. Key contributions include learning dynamic manipulation policies without expert demonstrations, leveraging low-cost, unmarked visual perception that tolerates low-frequency noisy inputs, and achieving strong sim-to-real transferability. Experiments demonstrate that the method enables fast, smooth, and robust control in rope stabilization and trajectory tracking tasks, generalizing effectively to unseen rope lengths, mass distributions, initial configurations, and external disturbances.

Technology Category

• Intelligent Robots: Manipulation
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Natural Language Processing: Safety and Robustness

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We address dynamic manipulation of deformable linear objects by presenting SPiD, a physics-informed self-supervised learning framework that couples an accurate deformable object model with an augmented self-supervised training strategy. On the modeling side, we extend a mass-spring model to more accurately capture object dynamics while remaining lightweight enough for high-throughput rollouts during self-supervised learning. On the learning side, we train a neural controller using a task-oriented cost, enabling end-to-end optimization through interaction with the differentiable object model. In addition, we propose a self-supervised DAgger variant that detects distribution shift during deployment and performs offline self-correction to further enhance robustness without expert supervision. We evaluate our method primarily on the rope stabilization task, where a robot must bring a swinging rope to rest as quickly and smoothly as possible. Extensive experiments in both simulation and the real world demonstrate that the proposed controller achieves fast and smooth rope stabilization, generalizing across unseen initial states, rope lengths, masses, non-uniform mass distributions, and external disturbances. Additionally, we develop an affordable markerless rope perception method and demonstrate that our controller maintains performance with noisy and low-frequency state updates. Furthermore, we demonstrate the generality of the framework by extending it to the rope trajectory tracking task. Overall, SPiD offers a data-efficient, robust, and physically grounded framework for dynamic manipulation of deformable linear objects, featuring strong sim-to-real generalization.

Problem

• deformable linear objects
• dynamic manipulation
• rope stabilization
• self-supervised learning
• non-negligible dynamics

Innovation

• physics-informed learning
• self-supervised manipulation
• deformable object dynamics
• differentiable simulation
• distribution shift correction