Topological Map Building Based on GNG Network in Dynamic Environment

Dynamic obstacles may result in inconsistent environment map. Aiming at this problem, a new building method of topological map is designed and realized for mobile robot. Firstly, the feasible region′s information is obtained via a probabilistic method to filtrate spurious measurements of dynamic obstacles. And then, with this information as the growing neural gas (GNG) algorithm′s inputs, a consistent topological map is built by learning and adding new node. This proposed method is selflearning and adaptive. Both the simulation and physical experiments verify the feasibility and effectiveness of the proposed method.