Controle planar para a movimentação de um robô quadrúpede

Abstract

Animals have the ability to move in unstructured terrain. In robotics, the ability to perform tasks in industrial, mining, oil and disaster environments is studied. However, for a robot to locomote efficiently in different environments, it is necessary to plan the steps as well as to control its legs in a synchronized way. This dissertation considers a quadruped robot with legs modeled using Euler-Lagrange equations. The proposed control strategy assumes an impedance control for each leg, which receives reference trajectories from gait patterns through 6th degree Bézier curves. The Bézier curves are constructed based on leg velocities through a planar kinematic model of the robot body. Finally, a body control allows the robot to follow a defined trajectory. Leg control, gait patterns and body control were implemented in Matlab and simulated in CoopeliaSim using the realistic Vortex physics engine. The results obtained in the simulations demonstrate that the proposed leg controller was able to track the Bézier curve used as a reference, making the robot move with static walking and trotting gait patterns, while presenting compliance to forces applied to the feet. The body control was able to command the robot to follow a reference trajectory, defined in this case by a Bernoulli Lemniscate curve.