3D Object Tracking using Growing Neural Gas with Different Topologies

Abstract

3D perception system of an autonomous robot is one of the most important technologies for performing tasks in an unknown environment. This paper proposes a real-time 3D object tracking method based on Growing Neural Gas with Different Topologies (GNG-DT). First of all, we explain a GNG-DT for learning the topological structures from the 3D point cloud data with multiple properties such as color and normal vector information. Next, we propose a real-time 3D object tracking method from the topological structure by utilizing the center of gravity in each cluster. Finally, we show some experimental results of the proposed method and discuss the effectiveness of the proposed method.