Time-Series Data-Driven Three Dimensional Shape Control of Deformable Linear Objects Using a Dual-Arm Robot with Dynamic Model Updating
Jiyoung Choi, Micheale Haileslassie Gebrezgiher, Donggun Lee, Ayoung Hong
AI summary
Problem
Controlling deformable linear objects (DLOs) like cables is challenging due to their high degrees of freedom, complex nonlinear dynamics, and the difficulty of obtaining accurate physical parameters for traditional modeling.
Approach
The authors propose a data-driven framework that uses a bi-LSTM network to predict DLO states from time-series position data, integrated with Model Predictive Control (MPC) and real-time online learning to adapt to varying object configurations.
Key results
- Time-series bi-LSTM model achieves higher prediction accuracy and faster inference than chain-like baselines
- Online residual learning enables real-time adaptation to dynamic environmental changes and material variability
- Input interpolation allows control of DLOs with different node counts without full model retraining
- Successful 3D shape control of diverse cables demonstrated in both simulation and real-world dual-arm experiments
Why it matters
Provides a practical, adaptable control framework for robotic manipulation of cables and sutures in manufacturing and surgical applications.
Abstract
Deformable objects (DOs) are prevalent in everyday environments and represent important targets for robotic ma- nipulation. However, their high degrees of freedom and complex nonlinear deformations make them more challenging to model and control than rigid objects when relying on traditional analyti- cal approaches. To address this, we propose a data-driven method to model the dynamics of deformable objects. Our method utilizes time-series data to predict future states without relying on complex dynamics. We employ model predictive control (MPC) for robot manipulation and improve its performance through online updates of the data-driven model. To handle cables with varying configurations, interpolation is applied to align model input structures. In this study, we focus on manipulat- ing deformable linear objects (DLOs) with different mechanical properties and configurations using a dual-arm robotic system, both in simulation and in real-world environments.