SBIR Phase II: Implementation of Machine Learning Module in Novel Relay Trucking Pilot

The broader/commercial impact of this Small Business Innovation Research (SBIR) Phase II project will be to enhance traditional U.S. long-haul trucking by eliminating costly downtown in supply chains, while making significant environmental and societal impacts. Currently, traditional long-haul freight trucking is limited to conventional point-to-point trucking models that require excessive idling, resulting in over $3 billion annually in unnecessary fuel and maintenance costs. More importantly, these inefficiencies contribute to mental and physical strains on truck drivers, exacerbating the industry's sustainability issues. With trucking demand projected to increase by 36% by 2031, this project's relay model aims to shift the status quo by enhancing asset utilization, effectively reducing delivery times by 20-50%, while lowering truck driver turnover costs. Transforming trucking into a local day job will significantly improve working conditions, while secondarily solving the driver retention and shortage crisis. Moreover, the relay model promises environmental benefits by cutting emissions from idling and empty backhauls while facilitating the adoption of battery-powered fleets. This Small Business Innovation Research (SBIR) Phase II project will build upon the machine learning (ML) based module software component developed in Phase I by validating its ability to quantify impacts of disruption events in long-haul relay trucking, resulting in a thorough and timely recommendation of mitigation strategies. Academic researchers have used simulation, mathematical programming, and other modeling techniques to establish the theoretical viability of trucking relay systems to solve equipment and human capacity issues; however, these models have relied on simplifying assumptions and do not account for common disruption events that pose a significant operational challenge. Quantifying the impacts of potential disruptions on travel time reliability while recommending timely and effective mitigation strategies to dynamically adjust driver schedules is essential to real-world deployment. Therefore, Phase II centers on four key objectives: 1) revising and integrating the ML models with real-time data stream APIs; 2) testing the scheduling engine (with integrated ML models) in a simulation environment; 3) piloting the software platform with live trucks and drivers on the road; and 4) analyzing key findings, incorporating changes, and creating a final report and revised product roadmap. Project learnings will translate to a practical relay software platform to propel commercialization. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.