| Resumo : |
In mobile robotics, autonomy is obtained by navigating an environment while using embedded sensors to estimate a map and the robot position with respect to this map. This problem is known as Simultaneous Localization and Mapping (SLAM) and has been extensively addressed for the single-robot case in the last few decades. However, when the environment becomes larger, a multi-robot solution for the SLAM problem is more efficient despite the increased complexity and the development of new problems, such as multi-robot coordination. This work aims to develop and implement a solution for the SLAM problem using multiple ground mobile robots. To accomplish that, each robot must autonomously explore part of the environment and their individual estimated maps must be merged into a single global map. The proposed solution uses a scan matching-based SLAM framework known as GraphSLAM to estimate pose and map for each individual robot. To autonomously explore the environment, the solution proposes the use of map segmentation via Voronoi Graphs, which generate collision-free paths and possible target points for exploration. Map merging is performed when robots rendezvous to take inter-robot measurements to provide relative position information. In order to validate the proposed approach, experiments using simulated and real robots were carried out to autonomously explore and map indoor environments. Real world experiments presented satisfactory results for a laboratory environment and a unstructured corridor environment and the proposed rules for coordination as well as the map merging approach worked as intended. |