Optimization of swimming locomotion for fish robots with multi-actuation

Enhancing the energy efficiency and swimming speed has long been a focused issue in the research of biomimetic fish robots. Existing methods are normally based on the offline analysis due to the lack of the closed-loop swimming control. In this paper, an on-line locomotion optimization approach will be presented to tackle the issue. The implementation of the approach comprises three parts: the on-line swimming gait generation based on artificial Central Pattern Generators (CPGs), the finding of optimal controlling parameters of the CPG model by using Genetic Algorithm (GA), and the construction of a closed-loop swimming control system based on feedback of swimming speed and energy efficiency. The proposed approach has been tested on an eight-DOF biomimetic propulsor, which mimics the anguilliform swimming. The design of the propulsor prototype, implementation of CPG and GA, measurement of the energy efficiency, and experimental results are presented.