Memetic search for vehicle routing with simultaneous pickup-delivery and time windows

The Vehicle Routing Problem with Simultaneous Pickup-Delivery and Time Windows (VRPSPDTW) has attracted much research interest in the last decade, due to its wide application modern logistics. Since VRPSPDTW is NP-hard exact methods are only applicable small-scale instances, heuristics meta-heuristics commonly adopted. In this paper we propose a novel Memetic Algorithm efficienT local search Extended neighborhood, dubbed MATE, solve problem. Compared existing algorithms, advantages of MATE lie two aspects. First, it capable more effectively exploring space, initialization procedure, crossover large-step-size operators. Second, also efficient exploitation, sophisticated constant-time-complexity move evaluation mechanism. Experimental results on public benchmarks show that outperforms all state-of-the-art notably, finds new best-known solutions 12 instances (65 total). Moreover, comprehensive ablation study conducted effectiveness components integrated MATE. Finally, benchmark large-scale derived from real-world JD logistics, introduced, which can serve as challenging test set for future research.