- The paper demonstrates that collaborative vehicle routing can achieve 20-30% cost savings and lower CO2 emissions compared to non-cooperative approaches.
- It systematically categorizes methodologies into centralized planning, decentralized planning without auctions, and auction-based decentralized planning.
- The survey outlines future research directions, emphasizing standardized datasets and strategies to address uncertainties and strategic behavior.
Collaborative Vehicle Routing: A Survey
The paper "Collaborative Vehicle Routing: A Survey" by Margaretha Gansterer and Richard F. Hartl provides a structured overview of the domain of collaborative vehicle routing (CVR) within logistics operations. This domain has gained significant traction given the potential efficiencies and sustainability benefits it can provide through the formation of coalitions among carriers.
The research identifies three significant streams in CVR: centralized collaborative planning, decentralized planning without auctions, and auction-based decentralized planning. It systematically reviews each stream, evaluating the state of current knowledge, proposing classifications, and suggesting future research directions.
Centralized Collaborative Planning
In centralized collaborative planning, a central authority with full information coordinates logistics operations. The literature is split into two main areas: assessing the potential benefits of centralized collaboration versus non-cooperative approaches, and developing innovative models or solution methods for such centralized planning.
- Collaboration Gain Assessment (CGA): Studies indicate that collaborative vehicle routing can lead to improvements in efficiency, with potential cost savings typically ranging from 20-30% when compared to non-cooperative solutions. This stream also highlights the ecological benefits, such as reduced CO2 emissions, particularly when collaborating for shared urban logistics.
- Methodological Contributions (MC): This area explores models for centralized planning, addressing the complexities of large-scale optimization problems. Techniques such as adaptive large neighborhood search and other decomposition strategies have been applied to solve challenges in VRP, ARP, and MCFP under centralized settings.
Decentralized Planning Without Auctions
Decentralized planning emerges when players are unwilling or unable to share full information with a central planner. This approach includes:
- Partner Selection (PS): Selecting suitable collaboration partners is crucial as potential benefits largely depend on compatibility, particularly their operational characteristics and requirements.
- Request Selection (RS): Carriers decide which requests to offer in partnerships. This selection is critical for optimizing collaboration benefits and is generally modeled through problems like the team orienteering problem.
- Request Exchange (RE): In non-auction-based settings, request exchange involves strategies to trade parts of existing routes rather than individual requests due to complexity constraints. Studies suggest iterative feedback mechanisms to improve collaboration outcomes.
Auction-Based Decentralized Planning
In this setting, auctions facilitate the exchange of requests among carriers, allowing them to serve as both buyers and sellers within a managed network. Combinatorial auctions receive particular focus given their ability to bundle requests effectively, which is advantageous in minimizing the risks associated with partial fulfillment.
- Multi-Phased Auction Processes: The survey articulates the various phases of auction processes, from request submission to bid allocation and profit sharing. It identifies the potential of auctions to reveal carriers' preferences indirectly and the complexities involved in achieving optimal solutions.
- Strategies for Bundle Generation: Recent research emphasizes the importance of strategic request selection and bundle generation, using methods like genetic algorithms to reduce complexity and computational effort without compromising solution quality.
Implications and Future Directions
The survey emphasizes the practical and theoretical implications of collaborative vehicle routing in logistics. It notes a significant increase in publications in recent years, reflecting growing interest and recognition of collaboration's potential benefits. Future research directions include:
- Addressing uncertainties in centralized planning and assessing collaboration gains in complex multi-modal systems.
- Investigating the impact of strategic behavior in decentralized planning and the integration of effective profit-sharing mechanisms to mitigate such behavior.
- Comparing auction-based to non-auction approaches to establish their relative advantages and disadvantages.
- Estimating the value of information within decentralized systems to enhance collaborative profits.
The paper concludes by calling for standardized datasets to facilitate comparison across studies, thus supporting the development of benchmarks within this research domain. Overall, the paper provides a comprehensive overview and forward-looking insights into collaborative vehicle routing, outlining the critical challenges and opportunities in this evolving field.
