Real-time moving objects tracking for mobile-robots using motion information

The perceptional of the motion of objects is a key problem for a mobile robot to perform tasks in a dynamic environment. Thus, we present a real-time approach for tracking multiple moving objects. The proposed algorithm initially detects moving regions and a dense optical flow technique is exclusively applied to those regions between two consecutive frames. Afterwards, the moving objects in each region are determined based on the planar parallax motion by assuming that independently moving objects undergo pure translation. For subsequent frames, the detected moving objects are tracked based on the orientation of the flow fields, while the new position is updated. In turn, the new detected objects are modeled during a tracking period. The proposed algorithm has been tested with various scenarios and the experimental results demonstrate that the proposed algorithm works properly. It can be shown that there is a significant reduction in the overall processing time for detecting and tracking multiple moving objects in a scene.