|A novel algorithm for multi-robot simultaneous localization and mapping is investigated through simulation. Although cooperative-SLAM has been studied by other researchers, the algorithm presented here offers a simpler, less computationally-intensive solution to map merging. The novel way to create local maps is identifying standard landmarks appearing naturally in the given area. The simulation results show the effects of three parameters: (1) number of robots, (2) map overlap matching threshold, and (3) map overlap window size. The success metric is the number of simulation steps required to fully explore a given map. This in turn validates the efficiency and accuracy of the map merging technique used.
The map overlap matching threshold establishes the degree to which compared parts of maps from two different robots must agree for the algorithm to join the two maps. The map overlap window size establishes the number of map cells that are considered when comparing maps from two different robots.
The results show that, within the range of parameters studied, increasing the number of robots significantly decreases the number of steps required to fully explore and map the given area. It is also shown that the number of steps required to fully explore the map increases when the map overlap matching threshold is increased. In the latter case, the simulation time also increases with increasing threshold.