Virtual functional segmentation of snake robots for perception-driven obstacle-aided locomotion?

Snake robots equipped with sensors and tools could potentially contribute to applications such as fire-fighting, industrial inspection, search-and-rescue and more. Such capabilities would require that a snake robot has a high degree of awareness of its surroundings (i.e. perception-driven locomotion) and is able to exploit objects and irregularities in its environment to gain propulsion (i.e. obstacle-aided locomotion). In this work, a simplified snake robot model is proposed to deal with a lower-dimensional system that allows for establishing the foundation elements of perception-driven obstacle-aided locomotion. To achieve this, a virtual partitioning of the snake into parametrised virtual functional segments (VFS) is presented based on the concept of virtual constraints (VC). The snake robot body is approximated by using a chain of continuous curves with the fewest possible parameters. These parameters can be treated as degrees of freedom of a “constrained system” and consequently be subjected to modelling and control at a higher abstraction level. The main contribution of the proposed conceptual approach is that the robot joint space can be reduced into a lower-dimensional space for articulation. This concept replaces the analysis of the individual mechanical degrees of freedom of the snake by the analysis of the functional roles of the parametrised VFS. The VFS are defined in relation to task space coordinates, and the roles of the physical links and joints change according to a defined set of transition events as they move along the robot's path. This method is a preliminary step towards realising perception-driven obstacle-aided locomotion for snake robots.