Certifiably Safe Manipulation of Deformable Linear Objects via Joint Shape and Tension Prediction

TL;DR

This work tackles safe manipulation of deformable linear objects in contact rich settings by jointly predicting future shape and tension and enforcing safety through a reachability based online trajectory optimizer using polynomial zonotopes. The method integrates a discretized DLO model with an LSTM predictor and a contact handler, a standard robot dynamic model with a tension driven interaction, and a robust RNEA based controller. Key contributions include a certifiably safe planning framework that enforces a tension bound with an error margin and a trajectory optimization that accounts for DLO deformations and contact interactions, validated in a simulated wire harness task where it achieves high success with zero safety violations. The approach improves safety and robustness in practical DLO manipulation tasks such as wire harness assembly, with potential impact on automation in assembly lines and other contact rich DLO manipulation scenarios.

Abstract

Manipulating deformable linear objects (DLOs) is challenging due to their complex dynamics and the need for safe interaction in contact-rich environments. Most existing models focus on shape prediction alone and fail to account for contact and tension constraints, which can lead to damage to both the DLO and the robot. In this work, we propose a certifiably safe motion planning and control framework for DLO manipulation. At the core of our method is a predictive model that jointly estimates the DLO's future shape and tension. These predictions are integrated into a real-time trajectory optimizer based on polynomial zonotopes, allowing us to enforce safety constraints throughout the execution. We evaluate our framework on a simulated wire harness assembly task using a 7-DOF robotic arm. Compared to state-of-the-art methods, our approach achieves a higher task success rate while avoiding all safety violations. The results demonstrate that our method enables robust and safe DLO manipulation in contact-rich environments.

Figures (4)

• Figure 1: This paper introduces a certifiably safe framework for manipulating DLOs in contact-rich environments. At the core of our method is a predictive model that simultaneously estimates the future shape and tension of the DLO. The figure illustrates a robot arm manipulating a DLO (white) toward a goal configuration (green) in the presence of an obstacle (red). The manipulation is executed without causing collisions or overstretching, despite contact with the environment.
• Figure 2: Overview of the proposed framework. Starting from an initial state, the robot generates a set of parameterized trajectories and uses a learned predictive model to estimate the resulting DLO shape and tension. These predictions are incorporated into a reachable set computation, enabling the identification of safe trajectories under contact-rich conditions. An optimization problem is then solved to select the best feasible trajectory, which is executed using a robust controller. The loop runs in a receding horizon fashion until the DLO reaches the desired goal configuration without collisions or overstretching.
• Figure 3: a) is an overview of the discretized configuration of DLO. Blue circles represent the discretized nodes of the DLO. b) shows the local coordinate of each node. For the $i^{\textrm{th}}$ node, its local coordinate is denoted by $(\alpha_{i}, \beta_{i}, \gamma_{i})$.
• Figure 4: Architecture of the DLO shape and tension prediction model.