Bayesian Optimization Based Terrestrial Gait Tuning for a 12-DOF Alligator-Inspired Robot With Active Body Undulation

Abstract

In addition to aiding in swimming, body undulation of an alligator plays a critical role in terrestrial locomotion by imparting stability. This paper reports design, fabrication and terrestrial locomotion control incorporating active body undulation of a 12-DOF alligator-inspired robot. Each of the four legs of the developed robot has two rotational degrees of freedom while the body can perform undulation using additional four rotational degrees of freedom. This paper also presents a Bayesian optimization based approach to tune the gait parameters of both leg oscillation and body undulation in order to maximize the average robot speed. We obtained improvement by a factor of 1.93 in average robot speed in comparison to the one obtained by randomly generated parameters and report the experimental results in this paper. In future, we plan to generalize the developed Bayesian optimization based parameter tuning approach for the swimming gait and thereby impart amphibious capabilities to the developed robot.