An improved algorithm for motion graph and its application in intelligent obstacle avoidance

Abstract

for a deficiency of classic algorithm in constructing motion graph, we propose an improved algorithm and further apply motion graph to the intelligent obstacle avoidance of virtual human. First of all, we introduce an approach for acquiring real human motion data by our active optical motion capture system and solving some related problems we have encountered. Secondly, we present and analyze the classic algorithm in constructing motion graph, and propose an improved algorithm to enhance its low computational efficiency, which accelerates the construction of motion graph. Finally, we apply motion graph to intelligent obstacle avoidance; we propose a novel algorithm to generate a suitable path, avoiding obstacles; then we use the classic motion graph search algorithm to synthesize a matched motion, driving the virtual human to walk along the planned path. Experiments we conducted verify the correctness and feasibility of the proposed algorithms.