In this paper we present an obstacle avoidance algorithm based on Particle Swarm Optimization (PSO), which is a stochastic optimization technique. The PSO algorithm was modified in order to solve the proposed problem, in our case each particle of the PSO represents a new position, during the PSO algorithm each particle is tested to see if it represents a valid position. The best PSO particle represents the new subgoal, this goal is used by a controller to move the robot from its current position to the subgoal. The algorithm is tested with different maps, that show the robot's path avoiding obstacles and reaching the goal.