Dynamic parcel-passenger matching optimization in public transit-based crowd-shipping system

Purpose Dynamic crowd-shipping systems based on public transit present unique challenges due to fixed schedules, stochastic passenger availability, and strict no-deviation constraints. This paper introduces the Dynamic Transit Crowd-Shipping Problem (DTCP), which focuses on making real-time parcel assignment decisions at transit departure events while accounting for operational feasibility, delivery timeliness, and uncertain future capacity. Design/methodology/approach We formulate DTCP as a sequential dynamic decision process triggered by scheduled bus departures, where parcels must be assigned to eligible passengers without altering their travel routes. To solve this problem, we propose a Probability-and-Time-driven Assignment (PTA) policy that directly approximates the policy function. PTA integrates immediate time cost, passenger availability, and residual delivery time into a unified decision framework. Extensive simulations are conducted across five system scales and multiple spatial alignment scenarios. Findings Results show that PTA significantly outperforms baseline strategies, including heuristic and value function approximation (VFA)-based approaches, in terms of delivery success rate and overall system utility. Notably, we find that the spatial alignment between parcel destinations and passenger flows is a critical determinant of performance. Although high passenger availability improves robustness, it cannot fully offset the negative impacts of spatial mismatch between supply and demand. Research limitations/implications This study extends real-time dynamic decision-making research to a previously underexplored domain of public transit-based crowd-shipping. It demonstrates the value of incorporating spatial-temporal structure and probabilistic reasoning into assignment strategies under strict route constraints. Practical implications The proposed PTA policy offers a practical tool for urban logistics planners and platform operators aiming to leverage public transit infrastructure for sustainable last-mile delivery. It enables efficient, real-time parcel-passenger matching without disrupting normal transit operations. Originality/value This research introduces a new class of crowd-shipping problems characterized by stochastic supply, fixed schedules, and zero detour tolerance. The PTA policy provides a novel decision framework that bridges operational research and real-world urban logistics needs, offering scalable and sustainable solutions for integrated transit-delivery systems.