Development of control strategies for autonomous navigation of multi-articulated robotic vehicles in backward movements

Abstract

This thesis is inserted in the area of mobile robotics, focusing on the problem of controlling wheeled terrestrial Multi-articulated Robotic Vehicles (MARVs). Such robots are generally characterized by a tractor element (active) and several trailers (passive). The aim is to provide these systems with the ability to perform navigation tasks, generally in environments with restrictions, such as obstacles that make it difficult for the robot to navigate in the environment. In this thesis, non-linear controllers were developed to guide these vehicles through certain path-following maneuvers. A strategy for transmitting velocities between the elements of the composition was proposed, which allowed the vehicle to be used without the need to get a complete model of the system. This strategy was validated and consolidated as one of the contributions of this work, as it is simple and applicable to any articulated chain. In addition, a dynamic control was implemented only for the tractor element, and it was possible to prove that this controller reduces navigation errors of the entire articulated chain. Another result of this research was obtained using adaptive control of the MARV, which made possible to navigate with small position errors, even when the load on the trailers changed during the stipulated route. Null-space-based control was also implemented to prioritize obstacle avoidance tasks, considering static and non-static obstacles. Another control structure tested was the heterogeneous formation of this MARV with an aerial robot for inspection tasks in environments with and without obstacles. The experiments carried out in this research were run using two trailers developed by this author and a commercial differential robot as a test platform. This thesis contributes to the literature for research in the area of navigation of multi-articulated mobile robots.