SELF-ADAPTIVE UNDERACTUATED HYBRID ROLLING/WALKING LOCOMOTION

In planetary exploration and other similar robotic applications, it is possible to encounter obstacles on multiple scales, making it difficult to design wheeled locomotion that works well for all terrain types. Legged locomotion tends to be less efficient and slower, but allows better obstacle clearance. This paper describes a novel method of achieving robotic locomotion over uneven terrain using a passive underactuation technique. Using planetary gear trains with one input degree of freedom and two output degrees of freedom, the natural obstacle-based locking of select outputs can cause the transition of power through the alternate outputs. By designing the primary outputs as wheels and the secondary outputs as legs with more ground clearance, a naturally adaptive hybrid gait incorporating both rolling and walking can be generated without the need for sophisticated sensing and control. Derivation and simulation validation are presented.