Abstract
In this paper, we describe a map acquiring system MARSHA for multi-robot environments. MARSHA addresses an important problem in multi-robot environments; how to share distributed/local knowledge. MARSHA tackles this problem in terms of relational expressions, which represent differences between every two robots. These relational expressions are computed by applying Matching Procedure to local maps which are generated by each robot individu-ally. By using these relational expressions, MARSHA makes it possible for all of the robots to utilize a global terrain map as their own map even if they do not travel around all the region. The effectiveness of the algorithm is shown through experiments with the implemented system on autonomous mobile robots.
