Abstract
Due to inherent compliance and non-linear behavior, modelling soft robots is highly complex. While the elasticity of their materials provides the robots with adaptability and resilience, it also causes undesirable effects. Cable-driven soft robots are particularly affected by frictional forces, which can significantly alter their deformed shape. Unfortunately, most existing cable-friction models have been developed for rigid surfaces and do not account for the interactions between elastic cables and surfaces. As a result, research on cable-driven soft robots often neglects friction due to the difficulty of acquiring data and implementing the mathematical models. This paper introduces a novel friction model that considers the asperity behavior of both the cable and the friction surface. We propose a new methodology that accurately replicates friction interaction in soft robots, which requires as few as nine data points. The methodology is assessed on four distinct material interactions, comprising two cables with different diameters and materials, and two 3D-printed surfaces made from PLA and TPU. By accurately estimating the nonuniform distribution of the joint deformation caused by friction, the new friction formulation achieves a 2.8% error in predicting a soft gripper's tip locations, while the current state-of-the-art model shows a 16.1% error. We also demonstrate that, with an accurate friction model, it is possible to optimize the cable routing points to achieve the desired grasping strategy for a soft gripper.
