A Novel Hybrid Map Based Global Path Planning Method

Autonomous navigation is a promising research domain due to its extensive applications. Path planning is the core of robot navigation. To improve the efficiency of path planning process in dynamic and complex environment is a crucial issue. In this paper, a novel global path planning framework based on a hybrid map is proposed. The presented map consists of three levels: the scene level, the feature level and the metric level. The environment is transformed into a graph of scenes and the scenes are defined according to the distribution of the environment features. In the path planning process, we first compute a global path based on the scene graph. Then the path is mapped into low levels to generate geometric path. As the high level is generated from the low level, the global path can provide strong guidance for computing metric path. The proposed method simulates the path planning process of human, which makes the most of the structure of the environment. The results of the experiments on the grid map demonstrate the effectiveness and high efficiency of our method.