Abstract: The problem of obstacle avoidance is important in multiagent flocking. Each agent should avoid obstacles safely， and then moves toward the target. Based on the existing artificial potential field flocking algorithm， an improved algorithm with obstacle avoidance capability was presented. In this algorithm， the obstacle was equivalent to a virtual agent for obstacle avoidance. Obstacle avoidance were not taken immediately when the agent perceived obstacles， but take the speed direction of the agent and the target point into consideration. According to different speed directions and position of the target， different obstacle avoidance measures will be taken. Through theoretical analysis and experimental verification， obstacles could be avoided efficiently based on the proposed algorithm， which can make the flocking faster.

Key words: multiagent, flocking, potential function, obstacle avoidance, tangent