Target behavior prediction method and device based on global path

The invention relates to a target behavior prediction method and device based on a global path, and the method comprises the steps: obtaining the motion state data of at least one target; generating a predicted track of at least one target according to the motion state data, and calculating an included angle difference between corresponding state points of the two tracks according to the predicted track and path data of a global path of the vehicle; and predicting the current behavior of each target according to the included angle difference, and predicting the future position of each target according to the current behavior. According to the method, the scene in the driving process can be recognized, the obstacle behavior in the scene can be predicted, it is guaranteed that the AVP vehicle makes an accurate decision on surrounding obstacles, collision is avoided, and therefore parking safety is guaranteed.