Fully distributed and decentralized map building for multi-robot exploration

Abstract

In many applications, a single robot may not be having capabilities to completely sense the environment in total due to several constraints like time and space. In such scenarios, each individual robot can sense a part of the environment and by sharing the same with others; each robot can build the information about the entire environment. This work focus on the problem of map building among robots in a team that enables them to sense things that are beyond their individual sensor range. The maps are built by individual robots and these maps are fused to generate a global map. The generated global map is then filtered to remove the redundant data. The memory used for this redundant data is thus freed and can be used to store other values. This map can be used in path planning or similar activities.