Intelligent path planning in 3D scene

This paper describes an improved version of D* algorithm and discusses how to apply this algorithm to intelligent path planning for virtual humans in complex and dynamic three-dimensional scene. Based on dividing the search space with navigation meshes, the improved D* algorithm can quickly calculate the optimal path from an starting position to the target position, and then guides virtual humans in walking in a dynamic 3D scene even though part of environmental information is unknown. At the same time, taking topographic differences into account, terrain costs are introduced into the heuristic function of D* algorithm, in order to get closer to the real effect of path finding. Feasibility and effectiveness of this algorithm are demonstrated by experiments.