SEMINAR: Systems in the Large: Analyzing Non-holonomic Motion from Robots to Animals by Dr Surya Singh

Abstract: Agile motion is so common that it seems deceivingly simple – one just accelerates, decelerates, or turns. However, combining these actions in a principled manner for optimal control is non-trivial as large and dynamic systems are coupled, compliant, and differentially constrained. More generally this can be viewed as an inverse control problem for non-holonomic systems. Analytic solutions to this are difficult to characterize. “Motion planning” treats this by transforming the continuous problem into a combinatorial one. This gives a more widely applicable solution that can additionally handle obstacle avoidance with computationally efficiency maintained by simultaneously considering estimates of likely motions and controls. This method is illustrated using examples from mining robotics (agile driving in unstructured environments) and biomechanics (gallop/running analysis).

About the Author Surya Singh gained his Ph.D. from Stanford University for a study on the locomotion of four-legged animals in which he attached inertial and image motion sensors to a dog in order to understand more about how dogs can run. This was part of a project to build a four-legged galloping robot in the Department of Mechanical Engineering at Stanford. Surya was with UWA Mechanical Engineering at 12 months as a Fulbright fellow, and then taught in the UWA school of Mechanical Engineering for 12 months before taking up his position at Sydney University.