Fuzzy control and dynamic simulation of a quadruped galloping machine

Motivated by the need for legged machines to use dynamically-stable running gaits in order to attain high speeds, this dissertation addresses issues in the simulation and control of a galloping quadruped machine. The gallop is studied because it is the preferred high-speed gait used by animals. A direct adaptive fuzzy control approach to the control of planar galloping and bounding is presented. The fuzzy controller is capable of learning the necessary leg touchdown angles and leg thrusts required for deadbeat control of the height and velocity of the quadruped at the top of flight. This is accomplished through an adaptation mechanism which is based on heuristics similar to those used in the controller designs of Raibert. Empirical relations concerning the galloping speeds of animals are interpreted in terms of the energy requirements of the galloping machine. The same fuzzy control approach is applied to starting and stopping the galloping machine in a single stride.; Efficient, general algorithms for the simulation of robotic systems containing kinematic loops are developed for use in evaluating prototype leg designs and in designing control systems. Two approaches, one based on acceleration constraints and the other on compliant loop-closure joints, are implemented and compared for a set of dynamic systems. Both approaches use Articulated-Body Inertia methods for simulation of the spanning tree of the system. Two methods for constraint stabilization of the acceleration constraints, Baumgarte and a spring and damper approach, are also implemented and compared. Finally, simulation and control results for a prototype leg, with a kinematic loop and in a one-legged hopper configuration, are presented.