Design and Experimental Validation of a Ballbot: A Self-balancing Omnidirectional Robot

This article presents the design, modelling and control of a Self-balancing Omnidirecional Robot, also known as ballbot, for performing autonomous navigation (path following) and object tracking in structured environments using vision feedback. A real ballbot has unstable and non-linear dynamics with second-order constraints, and is classified as an underactuated system. These features require the implementation of a robust and efficient embedded control system for allowing the realization of attitude and navigation control. In this context, the development of this work was divided into six stages: 1) design and mechanical construction of the omnidirectional wheels and chassis using a 3D printer; 2) mathematical modeling of the prototype developed in Matlab/Simulink and creation of a realistic simulated model in CoppeliaSim simulator; 3) design of LQR and P/PI controllers to control X and Y position and orientation respectively using the mathematical model; 4) evaluation of designed controllers using the realistic model developed; 5) implementation of the controllers in the embedded system and integration with inertial sensors and actuators; 6) experimental validation of the proposed control system for autonomous navigation and object tracking. The experimental results have shown that the developed ballbot is able to perform path following and object tracking efficiently.