Meeting Abstract
Snakes are extraordinarily capable of moving through cluttered habitats, tight spaces, and around obstacles, and consequently snake robots are a popular design for negotiating such situations. The remarkable locomotor performance of snakes in these environments has been attributed to both their elongate, serially metameric body plan and to their diversity of locomotor modes, each of which are used in different contexts. Concertina locomotion is the mode used by snakes to negotiate narrow spaces such as tunnels or bare branches, in which the snake forms an anterior anchor region while pulling the posterior body forwards, then forms a posterior anchor region while pushing the anterior body forwards. Although slow and energetically expensive, concertina locomotion is highly versatile, requiring only a surface to generate anchoring forces against, and is used as a final option when other modes are infeasible. In spite of the utility of this locomotor mode, concertina locomotion has rarely been implemented in snake robots, and only with a fixed, pre-specified tunnel diameter. In this paper, I use motion capture technology to conduct detailed examination of the anchor formation process in snakes using concertina locomotion in tunnels of two widths. These data were analyzed and used to design and implement an algorithm for a dynamic concertina locomotor mode in a snake robot, in which contact with the tunnel walls is automatically detected and used to modulate the waveform. This algorithm was applied to a “generalist” snake robot, allowing it to successfully perform concertina locomotion at a range of tunnel diameters by dynamically adapting its posture to the tunnel width based only on local feedback, with no prior information on tunnel width. This algorithm can be useful for snake robots when moving through tunnels, while limiting the need for either user input or complex sensor systems to characterize geometry.
