Comparing SONN Types for Efficient Robot Motion Planning in the Configuration Space

Abstract

Motion planning in the configuration space (C-space) induces benefits, such as smooth trajectories. It becomes more complex as the degrees of freedom (DOF) increase. This is due to the direct relation between the dimensionality of the search space and the DOF. Self-organizing neural networks (SONN) with their famous candidate, the Self-Organizing Map, have been proven to be useful tools for C-space reduction while preserving its underlying topology. In this work, the approach is adapted from human motion data towards robots’ kinematics. Furthermore, three additional models are implemented and compared to the existing ones with respect to their abilities to cover the used C-space of a robot and to preserve the topology. In total, the evaluation includes six SONN architectures, representing the consequent continuation of previous work. Generated Trajectories in the reduced SONN Output space were successfully tested in a robot simulation, providing a proof of concept for robot applications. The proposed method counteracts the Curse of Dimensionality for robots with many DOF, and thus a complete and optimal search algorithm can be used for path planning.