Obstacle Avoidance Path Planning and Motion Control for a Multi-Joint Soft Manipulator

Abstract

One of the most challenging tasks for soft manipulators is to avoid obstacles in a cluttered workspace. This paper aims to solve the problem by providing both an obstacle avoidance path planning method and a path following control strategy. To form a general framework for path planning, we proposed a collision map generation method in configuration space, which allows the usage of a general graph search or random sampling method to find a feasible obstacle-free path. In addition, to deal with the high nonlinear dynamics of soft manipulators, we design a gain-scheduling controller in the loop of pressure control, which adaptively tunes the controller gain along with the current operating pressures, and as a result, guarantees a unified performance for different deformation levels. The methods are testified on a three degree of freedom (DOF) soft manipulator and proven to be effective for obstacle avoidance tasks in cluttered environments.