Genetic Algorithm-Driven Inverse Kinematic Analysis of a Bio-Inspired Soft Robot in Discrete Space

Abstract This paper presents the inverse kinematic analysis of a planar bio-inspired soft robot in a discrete task space using artificial intelligence (AI). The soft robot manipulator comprises a stack of hydraulically amplified dielectric elastomer actuators, which exhibit snap-through instability, resulting in a discrete robot task space. The model simulates the manipulator with a series of adjustable-length rods connected with revolute joints and satisfies the piecewise constant curvature (PCC) requirement for each robot segment. The discrete task space of the robot was estimated using the Monte Carlo method. The inverse kinematic analysis was carried out through genetic algorithm optimization with an embedded forward kinematic model to minimize the error for both the position and the orientation of the robot's end-effector. We analyze the fully controlled robot at each discrete segment, and we compare the results with the reduced complexity model obtained by grouping the actuator in sets of different sizes. The algorithm showed excellent results and can potentially be used in the analysis of similar soft manipulators with discrete task spaces.