Yüksek taşıma kapasiteli dört bacaklı robot için sağlam yürüyüş kontrolü

This thesis presents a robust locomotion control framework for a quadruped robot. The locomotion control framework consisted of a trajectory planer and feedback controller based on CTC. Trajectory planning algorithm based on the ZMP concept to generate dynamically balanced locomotion and CTC method was implemented to the robot to follow generated trajectories. To implement the CTC, the inverse dynamic model of the robot is constructed through a recursive Newton-Euler method. Additionally, capture point method was implemented for the robot to preserve its balance against external perturbation. This method computes the capture point and generates a recovery trajectory for the robot to step in order to preserve its balance. Furthermore, novel algorithm is proposed to distribute contact forces for the purpose of eliminating undesired torso orientation fluctuations and regulate heading during dynamic quadruped trot-walking motion. The proposed method makes use of centroidal momentum, an important physical quantity in characterizing the whole-body behavior and overall balance of the robot. The contact forces are computed by means of centroidal momentum feedback and injected to the locomotion controller via Virtual Model Control (VMC) which can render virtual forces to regulate robot-environment interaction. The proposed controllers were validated via walking experiments in both a realistic simulation environment and by using an actual quadruped robot. Moreover, endurance and the payload capacity of the robot were validated through endurance test and walking experiment with $20~kg$ payload.