A study titled Efficient Neural Collaborative Search for Pickup and Delivery Problems by Kong Detian, a doctoral student from Research Center of Logistics at Nankai University, has recently been published in the prestigious journal IEEE Transactions on Pattern Analysis and Machine Intelligence (IEEE TPAMI), a leading publication in artificial intelligence (AI). Nankai University is the principal institution, with Professor Xiao Jianhua, Director of the Research Center of Logistics, serving as the corresponding author. The paper’s co-authors include Prof. Cao Zhiguang from Singapore Management University, Dr. Ma Yining from the National University of Singapore, and Prof. Yu Tianshu from The Chinese University of Hong Kong, Shenzhen.

IEEE TPAMI is widely recognized as a premier international journal in AI, with an annual acceptance number of around 200 papers. In 2024, it achieves an impact factor of 20.8 and ranks first in computer engineering, electronic engineering, and AI-related journals, with an h5-index of 165.

This paper introduces Neural Collaborative Search (NCS), a novel learning-based framework for efficiently solving pickup and delivery problems (PDPs). NCS pioneers the collaboration between the latest prevalent neural construction and neural improvement models, establishing a collaborative framework. NCS collaboratively trains the two models via reinforcement learning with an effective shared-critic mechanism. The construction model enhances the improvement model with high-quality initial solutions via curriculum learning, while the improvement model accelerates the convergence of the construction model through imitation learning. This paper also proposes the efficient Neural Neighborhood Search (N2S). N2S exploits a tailored Markov decision process formulation and two customized decoders for removing and then reinserting a pair of pickup-delivery nodes, thereby learning a ruin-repair search process for addressing the precedence constraints in PDPs efficiently. The N2S algorithm has demonstrated its diversity and efficiency through canonical PDP test sets, surpassing the renowned LKH3 solver on complex, constrained PDP variant.

Research Center of Logistics of Nankai University, is the only national logistics think tank co-established with the National Development and Reform Commission and the sole institution offering a doctoral program in Logistics in China. The center has developed a comprehensive logistics discipline system that integrates logistics economics, logistics management, and logistics engineering. This system fosters interdisciplinary collaboration, where the three fields are interwoven and mutually supportive, creating a distinctive logistics program.

For more details, the published paper is available at:

https://ieeexplore.ieee.org/document/10663917

（Edited and translated by Nankai News Team.)