Optimizing urban last mile delivery efficiency through dynamic vehicle routing heuristics and traffic flow analysis

Area of Science:

• Operations Research
• Transportation Science
• Computer Science

Background:

• Urban last-mile delivery faces challenges from traffic volatility, limited delivery windows, and disruptions.
• Static and single-strategy vehicle routing approaches are ineffective in dynamic urban environments.
• Existing dynamic VRP models offer partial solutions, lacking integrated adaptive search, congestion-aware costs, and proactive disruption management.

Purpose of the Study:

• To introduce a novel traffic-aware optimization framework for dynamic urban last-mile delivery.
• To integrate adaptive search, congestion-sensitive cost evaluation, and proactive disruption management into a unified system.
• To enhance the efficiency, reliability, and resilience of urban delivery operations.

Main Methods:

• Developed a Tabu-guided Adaptive Large Neighborhood Search with Rollout-based Real-Time Dispatch (T-ALNS-RRD).
• Enhanced the ALNS core with dynamic congestion-penalized cost functions.
• Implemented a multi-layered Tabu memory system for diversified exploration and a rollout-based dispatch for pre-emptive disruption response.

Main Results:

• T-ALNS-RRD achieved a 24.3% reduction in total operational cost and increased on-time delivery rates from 68.1% to 92.8%.
• Reduced congestion exposure by 54.4% and limited performance degradation under extreme traffic variability to 15.1%.
• Successfully handled 27.4 disruptive events per scenario with a 94.2% resolution rate and a 143.7 ms average response time.

Conclusions:

• The T-ALNS-RRD framework demonstrates significant performance advances for dynamic urban delivery optimization.
• The approach provides a scalable methodological basis for future large-instance deployments in logistics.
• Validated mid-scale framework offers improved efficiency, reliability, and resilience in complex urban delivery scenarios.