Task allocation fair-oriented heterogeneous motorcade goods taking and delivering vehicle path planning method

The invention provides a task allocation fair-oriented heterogeneous motorcade goods taking and delivering vehicle path planning method. In the first stage, a vehicle selection method with the shortest time consumption strategy is adopted. In a second stage, a novel model of parallel encoder-decoder structure, i.e., a heterogeneous attention model with parallel encoders, is proposed that fuses different attention mechanisms to automatically select orders to learn construction solutions intended to minimize the total travel time of the vehicles in the fleet. Experimental results show that the method is relatively fair in task allocation on most of data sets, time consumption of task completion is superior to that of other deep reinforcement learning methods and most traditional heuristic methods, and meanwhile, the method has obvious advantages in a large-scale order scene. According to the method, by means of route optimization, unnecessary travel avoiding and the like, the driving mileage and cost are reduce