The obstacle detection technique is one of the important part of perception module so that the obstacle avoidance capability could be achieved. In this research, the occurrence probability of obstacles in the global occupancy-grid map is proposed to help us identify the obstacles’ location. After the global-grid map is updated at each instance following the robot motion, the occurrence probability in each cell increases. Then, the obstacle location could be detected when the occurrence probability increases above a specified threshold. If the occurrence probability of each cell in the global-grid map exceeds a pre-defined threshold, the algorithm will label that individual cell as then obstacle instead of free space. In next iteration, the algorithm contin-uously updates until the robot reaches the target point. The obstacle position from this detection algorithm then passes to the potential filed method to generate the virtual force for the motion-planning system.