Obstacle movement prediction considering obstacle's dynamics and a priori knowledge of its goals

In this paper development of an advanced prediction method is proposed as a tool to enhance mobile robot navigation in dynamic environments. This research experiments the efficiency of this concept in the Small-size RoboCup environment in which the obstacles are the opponent soccer robots and, thus, not only move but act in an intelligent way to prevent our Multi-Agent System from achieving its goals. Understanding of the opponent's path-planning strategy and dynamical characteristics improves the accuracy of movement prediction. The focus is done in considering both the dynamics of the moving obstacles and a priori knowledge of their goals. Efficiency of this approach is tested for simple game configurations through simulations and experiments.