Reinforcement learning for node sequencing in transportation route design
Document Type
Conference Proceeding
Publication Date
1-1-2020
Abstract
Implementation of machine learning for the manufacturing sector can take advantage of its more efficient methods compared to traditional complex mathematical models. Exploration of the surroundings and collection of data is vital for an efficient route design in transportation of goods. In this research, a reinforcement learning (RL) approach is conducted to reconfigure the sequence of nodes to be visited by a road delivery transportation service if disturbances are present at any of the nodes. As the environment is explored, the RL algorithm collects data from its surroundings, performing an action in response and getting rewarded when a viable route is found and disturbances are avoided, learning more about its surrounding on every interaction. Heavy traffic or road closures can be considered as disturbances at the nearness of the node, therefore the sequence of nodes will be optimized to avoid visiting those with disturbances nearby, minimizing trip time, fuel costs, and vehicle utilization. Results of the proposed model show a faster convergence on a trained algorithm after including environmental details such as heavy traffic areas.
Publication Source (Journal or Book title)
Proceedings of the 2020 IISE Annual Conference
First Page
796
Last Page
800
Recommended Citation
Martinez, J., & Knapp, G. (2020). Reinforcement learning for node sequencing in transportation route design. Proceedings of the 2020 IISE Annual Conference, 796-800. Retrieved from https://repository.lsu.edu/mechanical_engineering_pubs/1466