Research Projects

Stop the Video

Research Projects

STATUS: In Progress YEAR: 2019 TOPIC AREA: Connected and autonomous systems Freight logistics and optimization CENTER: PSR

Dynamic Routing of Trucks and Truck Platoons Using Real-Time Traffic Simulators

Project Summary

Project number: PSR-19-04

Funding source: Caltrans

Contract number: 65A0674 TO 023

Funding amount: $100,000

Performance period: 1/1/2020 to 12/31/2020


Project description

Recent advances in sensing and navigation technologies makes it easier to route vehicles from origin to destination based on assumed traffic characteristics from historical data and available real time traffic data. Googlemaps and Waze are some of the most popular commercial applications used for routing instructions. These applications do not distinguish between different classes of vehicles and associated dynamics which often have a big impact on travel time and traffic flow characteristics. In areas where the volume of trucks is relatively high the impact of heterogeneous vehicle dynamics can be even more pronounced. This impact is expected to be worse if trucks began to get organized in platoons in order to save fuel and cut down labor cost. In addition, upcoming electric trucks in urban areas will add additional constraints for routing as well as impact on traffic flow. Developing mathematical models to be simple enough for decision making yet complicated enough to account for the dominant effects on traffic by new vehicle technologies is not an easy task. The availability of fast computers and advanced software tools enables the development of traffic simulators which can be used for more accurate routing decisions and closer optimality.

The purpose of this project is to research the feasibility of using real time traffic simulators for routing trucks and truck platoons both diesel and electric in a configuration with a route optimizer as shown in Figure 1. The traffic simulator is used to predict the traffic states based on historical traffic data as well as real time data from the network. Based on the predicted state information the optimization block generates the optimum routes which are evaluated using the simulation models in order to take into account the impact of the routed trucks on traffic flow states assumed in the optimization part and make the appropriate modifications.

The objective of the project is not to come up with new optimization techniques but rather use existing optimization tools and focus on how such tools can be integrated with a real-time simulator in order to improve truck routing. We will focus on a dynamic environment where incidents or other disruptions require the generation of new routes. What makes the problem challenging relative to past work is that we will address platoons of trucks whose size and dynamics cannot be simply modeled as just a ‘vehicle' like any other vehicle in traffic. The interactions of the truck platoon with other vehicles in the traffic network can be easily captured by the traffic simulator and taken into account by the optimization part. The traffic simulator will generate estimates of the link states and can be also used as predictor of link states ahead of time which will be used by the optimization to find optimum routes or update previous ones. The traffic simulator can also be used to evaluate such routes and the impact they have on traffic. For example, an optimum route for a truck platoon may have a negative impact on the rest of the traffic. The traffic simulator can be used to evaluate such impact by fast forwarding and mitigate the impact by making appropriate adjustments. Given that the bandwidth of decision making is relatively low i.e. decisions can be made over a couple of minutes long sampling intervals the simulator should be designed to have the time to carry out evaluations before the final decision is generated. We also plan to allow electric trucks and electric truck platoons to demonstrate how the traffic simulator can be reconfigured to account for their properties and how the optimization routing problem can be modified by adding the constraints of battery range, charging times and location of charging stations. In past work we considered the use of real time traffic simulators as part of a centralized coordinated multimodal freight load balancing, where we successfully showed the significance of traffic simulators in planning freight routes to achieve a good balance of freight loads across road and rail network. Due to the complexity of the problem, calculating the optimum distribution of all freight loads required a lot of time which makes the approach not useful in a dynamic environment where routing decisions need to be made fast enough. In this project we focus on a dynamic environment in traffic road network where decisions need to be made in real time and in addition we add upcoming vehicle technologies such as platoon of trucks and electric trucks.




Petros Ioannou
Professor of Electrical Engineering Systems, Ming Hsieh Department of Electrical Engineering; USC Viterbi School of Engineering
3740 McClintock Avenue
Hughes Aircraft Electrical Engineering Center (EEB) 200BLos Angeles, CA 90089-2562
United States
[email protected]