In logistics and supply chain management, there is a never-ending quest for continuous improvement, driven by the need for constant cost minimisation and service optimisation (Prado-Prado, 2009). During my bachelor thesis at a large European freight operator, I performed research into the optimisation of the firm’s (collaborative) road freight transportation planning. The case involved a select number of UK depots, each with a distinct planning office and a varying number of daily customer orders; pickups or deliveries. My proposed solution followed a ‘traditional’, mathematical, Linear Programming methodology. However, Generative Artificial Intelligence (GenAI) solutions with similar functionalities are rapidly advancing and becoming increasingly accessible. This caused the company to initiate a pilot that combined my solution and GenAI transport planning software, with its sights set on an autonomous vehicle fleet in the future.
GenAI-driven route optimisation solutions leverage machine learning algorithms, through the analysis of large volumes of historical and real-time data to create optimised routes, factoring in time, distance, and cost (OptimoRoute, 2023). Critical Success Factors (CSFs) and capabilities of these GenAI models include system integration, data availability, demand forecasting, adaptive routing, multi-objective optimisation, and flexible, multi-faceted constraints (FarEye, 2023). Through these CSFs, the primary benefits achieved by GenAI routing models relate to cost and CO2 emission reductions (Gangil, 2023). Other prevalent benefits include real-time adjustable planning and enhanced customer service, time efficiency, safety, and resource utilisation (FarEye, 2023; Patel, 2023). During my short time there, the pilot of my thesis company already elicited some of these benefits, primarily improved resource utilisation.
While there is great enthusiasm from practitioners and researchers surrounding these models (Abduljabbar et al., 2019), their limitations should not be overlooked. Most GenAI-powered transport planning solutions are still in early stages of development. Therefore, freight operators should be cautious of the accuracy of the generated routes, especially when the quality and quantity of the input data are questionable (OptimoRoute, 2023; Patel, 2023). Additionally, road freight transportation planning is a complex process, with a very broad range of relevant variables and actors. E.g., traffic conditions, vehicle characteristics, storage conditions, and delivery timeframes. GenAI models may not be able to factor in all these aspects, potentially resulting in incomplete or suboptimal routes (OptimoRoute, 2023). When carefully considering these limitations, GenAI transport planning modules seem to have tremendous potential, and, while not without challenges, they provide an exciting glimpse into the future of logistics.
Citations:
Abduljabbar, R., Dia, H., Liyanage, S., & Bagloee, S. A. (2019). Applications of artificial intelligence in transport: An overview. Sustainability, 11(1), 189.
FarEye. (2023). AI Route Optimization & Route Planning Guide: How AI Routing Can Transform Route Optimization. FarEye. https://fareye.com/resources/blogs/ai-route-optimization
Gangil, A. (2023). Generative AI-based route and logistics optimization. InfoSys. https://blogs.infosys.com/digital-experience/emerging-technologies/generative-ai-based-route.html
OptimoRoute. (2023). An Overview of Route Optimization Techniques. OptimoRoute. https://optimoroute.com/route-optimization-techniques/
Patel, R. (2023). AI Route Optimization: Does it Really Make Delivery Operations Efficient? UpperInc. https://www.upperinc.com/blog/ai-route-optimization/
Prado-Prado, J. C. (2009). Continuous improvement in the supply chain. Total Quality Management, 20(3), 301-309.