Robust Signal Control of Exit Lanes for Left-Turn Intersections with the Consideration of Traffic Fluctuation

Kaijia Chen, Jing Zhao, Victor L. Knoop, Xing Gao

Research output: Contribution to journalArticleScientificpeer-review

6 Citations (Scopus)
29 Downloads (Pure)


Traffic throughput at intersections can be improved by using exit lanes for left-turning (EFL), tidal flow lanes near an intersection, which has been recently introduced. This paper considers the operational robustness of the EFL intersection in relation to the control scheme applied to the traffic light setting. For safety and efficiency, it is important that the tidal lanes are emptied before traffic in the opposing direction uses these lanes. Hence, the signal control should not only be optimized for the mean value but be robust for all kinds of fluctuations. This paper formulates a traffic control scheme using robust optimization, i.e. an optimization scheme which explicitly accounts for extreme events. The fluctuation of traffic is considered from three aspects: the distribution of traffic demand, the distribution of base saturation flow rate, and the distribution of actual travel speed. Via a case study and extensive numerical analysis, we find that the established robust optimization method produces an efficient design of signal control and design speed at the EFL intersection under traffic demand and supply fluctuations. Though the optimization method is now applied to intersections with an EFL, it is considered useful for all intersections with high fluctuation of traffic demand and saturation flow rate.

Original languageEnglish
Article number9018258
Pages (from-to)42071-42081
Number of pages11
JournalIEEE Access
Publication statusPublished - 2020


  • Exit-lanes for left-turn intersection
  • robust optimization
  • signal control
  • unconventional intersections


Dive into the research topics of 'Robust Signal Control of Exit Lanes for Left-Turn Intersections with the Consideration of Traffic Fluctuation'. Together they form a unique fingerprint.

Cite this