Motion prediction of moving objects in a robot navigational environment using fuzzy-based decision tree approach

In a dynamic robot navigation system the robot has to avoid both static and dynamic objects on its way to destination. Predicting the next instance position of a moving object in a navigational environment is a critical issue as it involves uncertainty. This paper proposes a fuzzy rulebased motion prediction algorithm for predicting the next instance position of moving human motion patterns. Fuzzy rule base has been optimized by directional space approach and decision tree approach. The prediction algorithm is tested for real-life bench- marked human motion data sets and compared with existing motion prediction techniques. Results of the study indicate that the performance of the predictor is comparable to the existing prediction methods.