Incremental 3D Maps Server based on Feature Matching for Multi-robot Systems

This paper describes an incremental 3D maps server based on the point feature matching, which is dedicated to multi-robot systems. The idea behind the presented map server is that it maintains the 3D map by subsequent merging of received partial maps. The solution can be applied to distributed systems where each robot have its own map server instance and maintains a separate map instance. On the other side, in centralized robots systems, it can be one map server instance which integrates partial maps from robots, creates the common map and propagates it to the robots. The process of adding incoming maps consists of few steps. The first one is called a global alignment and is based on the feature extraction, description and matching with the assumption that the maps have an overlapping area. Then, the solution from the first step is used as the initial solution in the next step and it is corrected with a local alignment method. The last step is the data merging to obtain one, globally consistent model in the octree based representation. The presented solution was verified in the experiment with a group of wheeled robots - Turtlebots.