Title:DeRi-IGP: Manipulating Rigid Objects Using Deformable Objects via Iterative Grasp-Pull

Abstract:Heterogeneous systems manipulation, i.e., manipulating rigid objects via deformable (soft) objects, is an emerging field that remains in its early stages of research. Existing works in this field suffer from limited action and operational space, poor generalization ability, and expensive development. To address these challenges, we propose a universally applicable and effective moving primitive, Iterative Grasp-Pull (IGP), and a sample-based framework, DeRi-IGP, to solve the heterogeneous system manipulation task. The DeRi-IGP framework uses local onboard robots' RGBD sensors to observe the environment, comprising a soft-rigid body system. It then uses this information to iteratively grasp and pull a soft body (e.g., rope) to move the attached rigid body to a desired location. We evaluate the effectiveness of our framework in solving various heterogeneous manipulation tasks and compare its performance with several state-of-the-art baselines. The result shows that DeRi-IGP outperforms other methods by a significant margin. In addition, we also demonstrate the advantage of the large operational space of IGP in the long-distance object acquisition task within both simulated and real environments.