Geometrical Planner for Robotic Cable Routing in Cluttered Environments

The robotic manipulation of Deformable Linear Objects in cluttered environments is a challenging task. The main issue related to the robotic handling of these elements is their non-linear and distributed dynamics that results in a strongly underactuated robot-DLO system.As a consequence, extensive research has been made to model the behavior of DLOs, and a multitude of planning and control strategies based on these models have been implemented. However, the mathematical complexity of the models, the difficulty in estimating mechanical parameters, and the assumption of very low speed during manipulation to prevent dynamic effects from activating make these methodologies not suitable in real-world scenarios. Moreover, the interaction with environmental objects, which can improve the range of possible robotic applications, from knotting and wiring harness to cable shaping, remain underexplored due to the high level of complexity.Starting from the idea that stretching DLOs removes the nonlinearity issue, this work introduces a geometrical planner for tensioned cables to generate suitable paths for precise cable routing by exploiting environmental contacts while avoiding obstacles using only linear and circular motion primitives. Experiments conducted in different environments with different types of DLOs demonstrate the ability of the planner to precisely roll up the cable around obstacles, avoiding overstretching and collisions at the same time.Link to the video: https://youtu.be/-0zfLPSnw-E.