Snake robots are a class of hyper-redundant mechanisms that can traverse complex environments such as pipes and compact spaces with limbless locomotion, which are otherwise hard to access by legged, wheeled or flying robots. Despite such locomotion benefits, motion planning for hyper-redundant robots is difficult since one must coordinate all internal degrees of freedom (DoF) in robot’s shape space to achieve a desired net displacement in the workspace. To address the challenge, on the one hand, existing research has developed versatile gaits, cyclic motions in the shape space, to move the snake robot in desired ways. On the other hand, by combining these gaits with a state machine and switching between them as needed, locomotive reduction techniques simplify the control of a snake robot to that of a simpler system such as a differential-drive car, which can then navigate complex environments in a more intuitive way.
While such a combination of gaits and locomotive reduction is effective in many situations, the motion efficiency in terms of terrain traversal speeds of the robot is often limited. Most existing gaits for limbless locomotion rely on either pure undulatory motion or rolling motion of the snake robot, without an effective combination of both. On the one hand, the undulatory motion, such as sidewinding, allows the robot to continuously shift the contact segments along the backbone curve of the robot, and thereby transport itself by the friction at the varying contact segments. While being efficient and robust for various terrains, such undulation based gaits are still relatively slow to traverse even flat ground. On the other hand, relying solely on rolling, can also be slow for terrain traversal due to the small cross section of the robot and the torque limits of the joints.
We believe that relying solely on undulatory or rolling motion limits the locomotive capability of snake robots, and combining the two can enhance the speed of snake robot motion, without losing various gaits of snakes to locomote in complex environments. With that in mind, this project considers a mechanism which we refer to as a twistable snake robot (T-Snake), and focuses on its motion planning and locomotive reduction. T-Snake cannot only undulate its backbone curve as existing snake robots do, but also roll about its backbone curve infinitely to achieve wheel-like rolling when needed. Furthermore, different parts of T-Snake can roll with different angular velocities, resulting in twisting motion (and hence the name twistable snake). To plan motion for T-Snake, we introduce a rolling vector approach, which uses one or multiple rolling vectors to control the twist motion of the T-Snake, so that the robot can roll like wheels during undulation-based gaits such as sidewinding. Intuitively, when T-Snake is divided into two segments along its body, and two rolling vectors are employed, one for each segment, T-Snake can be locomotively reduced to a differential-drive car that can move forwards and turn in place. More detail can be found in our paper [1].
