Last-Mile Delivery Route Planning for Unmanned Vehicles Based on Improved Sparrow Search Algorithm

For the route planning of unmanned distribution in the last kilometer of the city, the matching between distribution and customer time is considered. Based on CVRP and constrained by narrow and soft time windows, the new model CVRP-NTW is proposed to optimize the multi-objective of minimizing total cost of distribution, minimizing total time spent, maximizing customer satisfaction and minimizing energy consumption. A new improved SSA algorithm is designed to solve the model, applying a reverse elite strategy in initializing the population and designing a sequence optimization strategy in decoding. Five different sizes of customer point instances are constructed for testing. GA and DQPSO are chosen as the comparison algorithms for ISSA. The experimental results show that ISSA can significantly improve the solution quality and solution speed.