Abstract: From the cognitive perspective,an approach by situation assessment technology was proposed for local path planning of mobile robot.First,the angle rang [10°,170°] in the front of mobile robot is divided into five parts under the robot coordinate system.The environmental situation factors of the five parts can be extracted for mobile robot from the fusion results of laser data and image data.Then,Bayesian networks model for robot action choosing is constructed.The environmental situation factors are regarded as evidences for the Bayesian networks mode.The inference can be made and the action in which the posterior probability is maximum is chosen.The action can be straight line walking,obstacle avoidance,escaping from U trap.Furthermore,the chosen action is processed to let robot move to the next grid.The next grid cell is chosen according to sonar data and the direction of robot is adjusted at the same time.Experiments including eleven typical simulation scenes were given.In these experiments,one scene test fails and the rest ten scenes are successful.In the ten scenes,mobile robot can reach the destination with the shortest route or secondary shortest route.The results show that the approach about situation assessment technology is effective and available for local path planning of mobile robot.

Key words: Bayesian networks model, Grid cell choosing, Local path planning, Mobile robot, Situation assessment