Physics-driven Locomotion Planning Method for Multilegged Robots

Abstract

Deformable robots have an excellent performance in unstructured terrain, because of the highly adaptive configuration. However, most current locomotion planning methods for deformable robots require elaborate manual intervention, so autonomy is urgently demanded. We propose a physics-driven locomotion planning method for multilegged robots. Using an artificial dynamic model, this method can produce the control command automatically once given the desired path. The artificial dynamic model includes only a few parameters with definite physical meanings. This method is evaluated on a 32-legged spherical robot using MSC ADAMS (Automated Dynamic Analysis of Mechanical Systems). The results show that our method is capable of generating a feasible locomotion control scheme for multilegged robots to track smooth paths.