Research Projects

Stop the Video

Research Projects

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

Robust Design, Analysis and Evaluation of Variable Speed Limit Control in a Connected Environment with Uncertainties

Project Summary

Project number: NCST-19-05
Funding Source: Caltrans
Contract number: 69A3551747114 TO 022
Funding Amount: $100,000
Start and End Dates: January 1, 2020 to December 31, 2020

Project Description

Connected vehicles via vehicle to vehicle (V2V) as well as vehicle to infrastructure (V2I) communications will open the way to manage and control traffic in a much more effective way than in today's traffic where sensing and control actions are very limited. While vehicle technologies are moving faster it is just a matter of time for the infrastructure to follow and provide the necessary instrumentation for V2I communication. The technology of V2I has been around for several decades and has been tested in automatic toll collection and other applications so one of the main remaining issues is the cost of investment and decision making. In order to properly evaluate the cost however, the stakeholders need to understand the potential benefits. The V2I technologies will allow more effective ways of dealing with incidents, controlling traffic flow and routing vehicles away from congested areas. Control techniques for traffic flow include variable speed limit (VSL), lane change control, ramp metering etc. In this project we plan to focus on VSL systems that are practical and whose performance can be established analytically and in simulations. 

Despite considerable research in the area of VSL control and the deployment of such technology in various places the full potential benefits of VSL as reported in the literature are controversial and often conflicting. The lack of consistency in the benefits of VSL may be attributed to various factors that include accuracy of the models used, lack of robustness of the VSL controllers developed in a way that a small disturbance may lead to considerable deterioration of performance, lack of rigorous analysis to support and explain simulation results etc. The purpose of this project is to address the issue of robustness in the design of VSL and bring such schemes closer to a successful implementation with consistent and well understood benefits. This is an extension of our past work where we identified all equilibrium states of the traffic flow under different demands, developed VSL schemes to force traffic flow to converge exponentially fast to the desired optimum equilibrium that corresponds to the maximum flow and lowest density under different initial density conditions and varying demand. In this past work we assumed the absence of any uncertainties in order to develop an ideal performance benchmark to compare with, under non ideal conditions. With the addition of uncertainties we will no longer have equilibrium points in the flow-density relation but equilibrium sets whose size may depend on the level of uncertainty, therefore even though the past work is useful to this project the results and analysis do not directly apply. We plan to use a traffic flow model that incorporates modeling uncertainties in terms of unmeasured unknown disturbances, parametric uncertainties in the various parameters of the model, density and flow measurement noise and uncertainties in the assumed fundamental diagram. With such more realistic model we plan to analyze its open loop properties to be consistent with what we observe in practice (qualitatively at least) and from microscopic simulations (more quantitatively). We then plan to use this understanding to design robust VSL (RVSL) schemes which will maintain acceptable performance in the presence of uncertainties. Unlike most of the work in the area of VSL our approach is to rigorously analyze all properties of the RVSL under different initial density conditions and varying demands as well as different uncertainties. Rigorous analysis is something that is missing in most of the work on VSL which often rely solely on simulations to show benefits. As a result it is difficult to understand inconsistencies and differences in results published by different researchers. In addition to analysis we plan to demonstrate our analytical results and benefits of the designed RVSL schemes using first macroscopic and then more realistic microscopic simulation models of the traffic on I-710 where the relative volume of trucks is high introducing more uncertainties in the macroscopic model, under different levels of demands and incidents. The EPA model MOVES will be used to evaluate the emissions and impact on environment with RVSL and without.


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]