Optimization of Stochastic Robotaxi Dispatch Problem with Boarding-Time Recommendation

This work investigates a stochastic robotaxi dispatch problem with boarding-time recommendation. Under an environment with stochastic travel time, a multi-objective stochastic programming model is formulated with the objectives of minimizing passenger experience penalty costs and minimizing total vehicle travel distance, allowing the platform to recommend appropriate boarding times to users. To solve this model efficiently, a clustering-based heuristic initialization method is designed according to the spatial distribution of requests to generate high-quality initial populations, iterated local search is employed to intensively explore promising solutions to improve algorithmic convergence, and a stochastic simulation method is introduced to evaluate the obtained solutions. On this basis, a multi-objective evolutionary algorithm with an iterated local search method (MOEA-ILSM) is proposed. Test instances are constructed based on real datasets from Didi Chuxing and the road network of Chengdu, and comparative experiments are conducted against three representative multi-objective optimization algorithms. The results demonstrate that the proposed algorithm outperforms the comparison algorithms in solution quality. Ablation experiments further validate the effectiveness of the boarding-time recommendation mechanism, showing that it can reduce passenger waiting time and improve the ride-sharing rate of requests.