Coordenação ótima de múltiplos robôs de serviço em tarefas persistentes

Abstract

Monitoring an environment using multi-autonomous robots operating in a shared workspace without collisions with each other is a great challenge nowadays. Many researchs have been done to develop this important task in robotics. Thus, this thesis deals with coordinating the motions of multiple working robots traversing periodic intersecting paths. It is desired that mobile robots following the initially designed paths without change them during the mission. To avoid any collision a centralized controller was developed to planner the velocity profile of the robots over the predefined paths, maximizing the smallest time margin t that the robots cross over a same collision point. The centralized solution used in this work was modeled as a Mixed Integer Linear Programming (MILP) problem using the Linear Interactive and General Optimizer (LINGO) software to maximize t and to find for each robot the sub-path traversal time. It was developed a C++ code capable to start the optimization process and receive for each robot the optimized traversal times, used to calculate their respectives average speeds. Besides, this code was used to control the robots, which was done through the Robot Operating System (ROS) framework integrated with LINGO. A computer vision ROS package named ARPOSE was used to localize the robots during the task, calculating their poses relative the global reference frame. Finally, to demonstrate the effectiveness of the approach, simulantions and real tests was performed on mobile robots group. The results obtained were satisfactory, which allowed to achieve the goals.