Transporte cooperativo de objeto por multi-robôs

Abstract

Coordinated transport of objects by a robotic swarm can be advantageous when the object to be transported is too large and/or too heavy to be effectively handled by a single robot. Cooperative transport requires coordination and synchronization of the pushing forces, to be exerted by the robots in order for the object to be transported successfully. In this work, a new algorithm for Cooperative Transport of Object by Pushing, called TCOI, is proposed. The algorithm has four stages. The first stage implements the object search done by the robots. The second stage of the algorithm starts when at least one robot finds the object. It then recruits the other robots. During the third stage, each robot first computes and then goes towards the location where it must position itself behind the object to cooperate in the pushing actions required for executing the transport properly. When the positioning of all robots behind the object is completed, the fourth stage of the algorithm begins. From then on, the robots start alternating between pushing and re-positioning actions in order to keep the object moving along the expected transport trajectory. This procedure is executed until the object is actually homed in a predefined location. In this work, the implementation of the object search stage is inspired by two strategies. The first strategy is the random search. The second strategy mimics the behavior used by ant colonies during their foraging for food sources, depositing pheromones along the traversed paths. Algorithm TCOI is implemented using a swarm of GRITSBot robots, via the Robotarium platform. In addition to the performance evaluation of the two search strategies, we investigate the impact of the positioning of the robots during the pushing actions, that of the number of robots as well as the impact of the diemnsion and weight of the object. To evaluate the proposed project, the time of each stage and the distance traveled by the object are evaluated. The results show that the search method, which mimics the behavior of ant colonies, obtains a shorter route than the method using random search. On average, the results using the behavior of ants show a reduction in the route of 6,7%, for moving robots forward only, as for the forward/backward movement, the reduction in the route is 52, 35%. In the analysis of the impact of the positioning of the robots, the maximum angle between the robots provides a smaller error in the object path length, than for the other angles. On average, the results show a reduction in the error in the length of the object’s path of 75, 76% between the maximum and minimum angle between the robots. In the analysis of the impact of the number of robots, it is possible to verify that the more robots that push the object, beyond the required minimum, the faster the object transport stage becomes. The reduction in time to transport the object using the maximum and minimum number of robots is 829, 77s. Finally, in the analysis of the impact of the object size and weight, it is confirmed that the minimum number of robots to push the object must be proportional to these values.