Probability map building of uncertain dynamic environments with indistinguishable obstacles

This paper proposes a method to build a probability map of the environment for navigation. It is assumed that the environment has multiple indistinguishable moving obstacles and the vehicle has limited sensor range and, therefore, lacks global information. The probability map is updated through the measurement and the probabilistic model of the obstacles. The model is obtained from a priori statistics of their movement. Probabilistic data association method is used to track multiple obstacles even after the vehicle loses tracking of some obstacles. The error bound of the algorithm is also analyzed.