Proposal Summary and Objectives

Freight traffic, particularly when it’s significant in proportion, affects the performance of the road network in a more sensitive and significant way compared to other type of traffic. It affects the aspects of mobility, environment, and safety due to the complexity of characteristics of the resulting mixed-class traffic. Trucks need extra distance and time for deceleration and acceleration, and their interactions with conventional vehicles can present more uncertainty to the traffic due to their lengths and speeds. Therefore, a traffic bottleneck appears more easily on a road segment or intersection where freight traffic is significant. Therefore, research insight into the control and operation of significant freight traffic is necessary. Previous FMRI research (year 1) has shown that the coordination of signals fails when the demand is composed of a large portion of trucks. Strategies have been developed in a FMRI second-year project to formulate multiple trucks’ trajectories to pass consecutive signals individually and cooperatively considering mixed traffic conditions. The stability problem of vehicle streams has been studied in the third-year project.

This exploratory research is proposed to operate vehicles (trucks) using artificial intelligence technologies. The objectives of the research are to improve the mobility of truck-car mixed traffic by saving average travel time, reducing emissions, and pollutions for trucks as much as possible though imitation learning, and the experience for the learning is from the results of optimized models we have developed so far. As a basic requirement, safety is the priority. The developed algorithm must ensure collision avoidance for all the vehicles. In addition to that basic requirement, other optimal performances are targeted. The expected output is a well-trained AI model and it will let trucks drive in behaviors that meet these requirements and performance objectives.

Funding Amount: $70,000
Status: Active
Duration: Sep 1, 2020 - Jul 30, 2021