Threat Cost Based Multi-level Prediction D-star Algorithm

Abstract

: In this paper, we propose a Multi-level Prediction D-star algorithm (MLP D-star) based on threat cost to address the path planning problem of mobile robots in local unknown environments. The algorithm improves the node expansion of the D-star algorithm using a multi-level prediction structure, which avoids excessive turning points in the planned path. The cost function of this algorithm incorporates threat cost and heuristic function to prevent the issue of path crossing obstacles. Simulation results demonstrate that the improved MLP D-star algorithm has advantages in terms of real-time performance, practicality of path results, safety, and computational efficiency.