Exploração e transporte cooperativo de objetos em ambientes desconhecidos

Abstract

This dissertation presents a framework for identification and transportation of objects arranged in an unknown environment. The objects shape, size and weight can be varied and transported in two ways: grasped by a manipulator coupled to the robot body or pushed. According to the robots capacity and the objects characteristics, a transporting mode is more appropriate than the other. It leads us to question which transporting mode should be performed and which robots should be allocated for the transport task.Most of the approaches that address the cooperative transport consider that robots have prior information of the object characteristics. In this work the robot must estimate the needed object characteristics for the transportation decision process. For that, the robot uses a laser to make a 2D object map. Furthermore, it estimates the effort required to push the object through measurements of the engine power. With this information the robot can decide how the object will be manipulated and request help from other robots to perform the transportation, as needed.We present simulation and experimental results to evaluate the framework efficacy and limitations. Although some assumptions limit the system, we came to a conclusion that the framework provides cooperation among the robots to carry out the proposed task.