LIDAR-based dynamic environment modeling and tracking using particles based occupancy grid

Due to the complexity of environment, it is a challenging task to detect and track moving objects in driving environment. In some simple scenarios, modelling and tracking of object can be realized by using the geometric properties of the object; In the more complex cases, it can be modelled and tracked as an occupancy grid. This paper presents an occupancy grid tracking system based on particles. The particles will have the dual nature - they will denote hypotheses, as in the filtering algorithm, but they will also be the building blocks of our modelled world. The particles have position and speed. They can migrate in the grid from cell to cell according to their motion model and motion parameters. The particles will also be created and destroyed through resample algorithm. The tracking algorithm is based on the particles, instead of cells. The occupancy grid tracking solution that proposed in this paper can be classified as using the Descartes probability model of the reverse sensor and generates a fully dynamic grid. This method is suitable for flexible and dynamic unstructured environment.