Gait Research

Snake Estimation

The snake robots' potential to locomote in a variety of terrains suggests a number of practical applications. In particular, we have been interested in applying snake robots to urban search and rescue, along with industrial inspection and other safety, security and response applications. The key to being able to locomote in all these environments is the robot's many internal degrees of freedom (DOF) of these mechanisms. These many DOF, however, pose deep fundamental research questions, including but not limited to mechanical design, control, and motion planning.

Snake Estimation

In terms of mechanism design, we have already built a family of 16-DOF Snake Robots Our design is modular; each module consists of a single rotational DOF performing low-level PID control to commanded angular positions. To produce motion in three dimensions, the axes of two adjacent modules are offset by 90 degrees.

Much of our lab's research is not in the robot itself, but rather its locomotion. To simplify control of the snake's many degrees of freedom, cyclic motions, known as gaits, have been developed that rely on pre-defined undulations that are passed through the length of the snake. In our case we use parameterized sine waves that are based on Hirose's serpenoid curve [Hirose: Biologically Inspired Robots], and its 3D extensions. Since our snake robots consist of modules where the joints are alternately oriented in the lateral and dorsal planes of the snake, our framework for gaits consists of separate parameterized sine waves that propagate through the lateral (even-numbered) and dorsal (odd-numbered) joints. We refer to this framework as the compound serpenoid curve.

These gaits enable an intuitive mapping between changes in a handful of parameters and motion of the robot through the environment. Over the years, our lab has developed gaits that can traverse a variety of terrains, including flat ground and pipes. Optimizing the parameters of this powerful low-dimensional framework and extending it include a richer of variety of motions is an area of ongoing research.

Our robots can be rapidly reconfigured into customized articulated legged or legless morphologies capable of mobile manipulation and inspection. We research creating behaviors for modular robots, with algorithms fast enough to match the pace of physical reconfiguration