Behavior-based Control of a Hybrid Quadruped Robot

Abstract

The mechanics and control specifications of a hybrid quadruped robot currently being built at Harbin Institute of Technology, China is proposed. The robot, named HIT-HYBTOR, weights about 25 kilograms, has four legs and wheels on its feet so that it can walk and step over barriers on rough terrain or drive by wheels accessing more speed on even ground. To master the control task of HIT-HYBTOR, hierarchical hardware architecture was implemented with two levels: the microprocessor level and the PC level. The microprocessor level is responsible for the low-level control of the legs, the wheels and the body movement, of course, receive the sensor data too. The PC level controls the complete movement of the HIT-HYBTOR like trajectory planning and leg coordination. A behavior-based control system clasified into three kinds: step behaviors, posture behaviors and navigation behaviors for the quadruped walking robot is described. We investigate how the behavior-based approach can be used for control and gait generation of a quadruped walking in uneven terrain. Results from simulations and test of the robot are given.