Graduation date: 2007
A distinct characteristic of legged locomotion is its periodic nature. This periodic motion, in the form of a periodic orbit, has been the target of many walking and running control strategies. The spring loaded inverted pendulum (SLIP) has become a popular model of sagittal plane locomotion, exhibiting behavior characteristic of a variety of legged animals. In this work, a model predictive control scheme is developed for the rigid body SLIP to drive the system to a periodic orbit. This is accomplished by defining a Poincaré map from one stride to the next and using numerical optimization each stride to select a leg touchdown angle that will best deliver the system to a desired fixed point of this map. The scheme is tested on both the point mass and rigid body SLIP models using parameter values that are characteristic of the cockroach, Blaberus discoidalis. It is found to increase the region of stability for both, as well as greatly improving the systems ability to recover from energy conservative perturbations.