A hierarchical pedestrians motion planning model for heterogeneous crowds simulation

The paper has proposed a hierarchical method for heterogeneous pedestrian motion planning in complex dynamic environments. It extends the Minimum link paths (MLPs) in a constraint-based algorithm to perform pedestrians' global path computations for avoiding diverse obstacles in complex dynamic scenes. Simultaneously, it uses a modified social force model to capture pedestrians' local dynamics with distinct individual characteristics and intermediate goals. The model has augmented the pedestrian's cognitive capability on environment. At global level, it minimizes the shortest global path graph for avoiding obstacles. At local level, it intelligently selects the immediate target in global path graph which instructs the pedestrian local motion, instead of using a pre-defined or a single goal attraction path potential. We simulate the model in a complex indoor scenario consisting of multiple pedestrians and diverse obstacles to generate realistic animation.