An approach to adaptive map building of mobile robots in unknown environment

Map building in unknown environment is the prerequisite to perform other tasks for the mobile robots. An effective approach based on the frontier points is proposed in this paper. Firstly, an adaptive clustering algorithm which integrates subtractive clustering and k-means clustering is introduced to partition the frontier points into different groups. Secondly, the centers of the groups serve as candidate destination points and they are evaluated using multi-step-ahead prediction. The optimal one will be selected to guide the robot to explore the environment with high accuracy. Finally, simulation results are provided for validation which show that the proposed strategy can generate accurate and complete (or nearly complete) maps of the unknown environments.