The objective of this paper is to design a quadruped robot with compliant legs. Compliant legs are developed using flexible joints, which allow the robot to attenuate the effect of support-leg exchange. A model of the legs is generated in SimMechanics and optimized to minimize the maximum torque required by the actuators. The control of the robot is divided into two stages: (ⅰ) a gait central pattern generator, and (ⅱ) a control system with feedback linearization. The gait pattern generator is developed based on optimal inverse kinematics with the use of Bezier polynomials. The resulting gait is used as a set-point in a closed-loop feedback control, which achieves a stable gait locomotion over rough, uncertain terrain. The uncertainty on the terrain causes unknown impacts in the robot. These impacts are absorbed by the compliance of the leg mechanisms. The proposed leg mechanisms are tested in a 3D-printed quadruped robot with fifteen degrees of freedom. Since, the robot is designed with eight actuators, the robot has seven degrees of under-actuation. The lack of actuators in this robot is overcame through the proposed gait pattern generator.
展开▼
