The overall objective of this project is to create computational algorithms for legged agility, which is defined as the robots ability to change its speed, direction, and/or gait (e.g., walk to run). As evident from biology, legs provide a highly agile mode of locomotion, yet no legged robot has been able to achieve performance comparable to highly performing humans or animals. The continued existence of this need is an important problem because it limits the usefulness of legged robots; the need to start, stop, change directions, accelerate/decelerate quickly are all essential if legged robots are to replace wheeled robots in rough terrain. Some applications of such agile robots are, as first responders, as search and rescue personnel, and as industrial workers. In addition, this research will provide training opportunities for multiple undergraduate students and outreach to school children including target populations that are traditionally under-represented in Science, Technology, Engineering, and Mathematics (STEM).
The key idea to achieve agility is to combine steady-state (periodic) gaits to achieve non-steady gaits. The rationale is that computation-wise, steady-state gaits are at least an order of magnitude cheaper to compute than non-steady gaits. The specific aims of this research are to create tools for (1) filling the entire state space with steady-state gait controllers and associated stabilizing controllers, (2) transitioning between steady-state gaits to create agile gaits and (3) experimental validation of the approach. A novel formulation of the Lyapunov function allows the creation of stabilizing controllers and estimating the regions of attraction (ROA). Then these ROAs will be composed sequentially based on their overlapping regions to create transition controllers for agile locomotion.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
