<p>The autonomous delivery robots (ADRs) are confronted with significant challenges, including mileage anxiety and limited payload capacity. This study intends to advance research on the electric van-based ADR delivery in urban areas. By integrating mobile charging technologies and photovoltaic storage charging facilities, it is possible to alleviate concerns related to charging and to reduce the indirect carbon emissions generated during the delivery process. Moreover, the implementation of split delivery for ADRs can more effectively satisfy the actual demands of customers. To address the electric van-based robot location-routing problem that considers mobile charging and split delivery, this research establishes a mixed integer programming model to minimize the daily operating cost of logistics enterprises. An improved shuffled frog leaping algorithm (ISFLA) is introduced, featuring a three-stage hybrid heuristic initialization and a modified local deep search method with multiple moving and shaking operators to enhance solution quality. Numerical experiments show that ISFLA outperforms the original version and the local search genetic algorithm in optimal solution quality, average performance, computation time, and stability. Experimental results confirm that the electric van-based ADR delivery model is more cost-effective than using only diesel vans. Finally, sensitivity analysis results reveal that the payload capacity and endurance mileage of electric vans and ADRs, as well as the adoption of mobile charging price, are key factors influencing the daily operating cost of logistics enterprises.</p>