Active suspension control for compliant wheeled locomotion on uneven terrain

Legged robots, as mobile systems, have demonstrated remarkable potential in navigating rough and uneven surfaces. Their multi-limb structure allows for a versatile interaction with the environment, coping with terrain disturbances and deformations. However, the motion complexity could arise due to this articulated design. The robot CENTAURO employed in this thesis presents a considerable mass and dimension that reduces its agility. The introduction of wheels will ensure a simpler and less energy-consuming locomotion method. Controlling these robots requires strategies that must take into account the unstructured nature of the terrain. Position-based controllers struggle to adapt quickly and accurately to the irregular profile of the environment. Furthermore, ensuring stability through criteria that rely on the definition of a support polygon becomes challenging in non-coplanar conditions. To expand a real-time framework developed in the laboratory, a control strategy is proposed for the lower body of the robot CENTAURO to execute wheeled locomotion in a real-life scenario. This strategy is composed of two controllers, addressing the compliance and stability of the robot on a non-coplanar surface. First, a Cartesian Impedance controller has been developed, such that the robot’s legs will act as an active suspension system able to quickly react in a compliant way against non-predicted changes in the terrain. Secondly, using the information coming from proprioceptive sensors and IMU measurement, an attitude controller was developed to ensure the robot maintains its base parallel to the ground plane. This functionality guarantees stability while traversing slopes with significant altitude differences. The controller has been vastly tested, both in simulation and in real-world trials with the robot. Experiments have shown the ability of the Cartesian impedance control to negotiate obstacles that might otherwise cause the robot to tumble with non-compliant behaviour. Future work will focus on implementing a more sophisticated strategy that will resolve some of the simplifications made in the robot's dynamics, aiming to ensure the stability of the robot.