Co-Optimization of Morphology and Actuation Parameters of Multi-Sectional FREEs for Trajectory Matching

Abstract

Fiber Reinforced Elastomeric Enclosures (FREEs) have gained significant popularity as a form of soft artificial muscle for the diversified deformation behaviors upon pressurization. In particular, modular FREEs connected in series, demonstrate enhanced flexibility and reconfigurability, thus are adaptable to more complex trajectory tasks. Morphology and motion parameters are strongly coupled for soft manipulators to achieve desired motion behaviors. In contrast to the traditional way of alternating between design optimization in either morphology or actuation space, we seek to co-optimize over parameters in both spaces in an end-to-end fashion with gradient-free optimization. The co-optimization framework allows for faster convergence towards optimal performance given desired trajectory plans. We demonstrated the feasibility of the framework with two simulated tasks: the multiple shape matching task and the octopus reaching task. The former task was further evaluated with real-world experiments.