Page 9 - Penn State Civil and Environmental Engineering 2021 Annual Report

FACULTY RESEARCH FACULTY RESEARCH CEE NEWSLETTER • VOLUME 37 2021
9
Gayah tested the hybrid method through a a a a a a a a a a simulated square network in fin in a a a a a a a a a a variety of scenarios finding that all three methods—PBIL Bayesian optimization and hybrid—identified configurations that led to more efficient traffic patterns than a a a a a a a layout with with zero restrictions However in in simulations with with more realistic settings the the hybrid method proved to be the the most effective According to to Gayah the most efficient configurations tended to to ban left turns in the the the the middle of of the the the the city and allowed them more often on the the the the the periphery While the the the the method was applied to a a a a a generalized network the the results can be used as a a a a a a a a a starting point for real-world traffic patterns with the algorithms being customizable on a a a a a a city-by-city basis “The grid network is the most generalizable and not not specific to any city ” Gayah said “I cannot take the best configuration for New York and apply it it to San Francisco but this generalized approach could be configured for any network with a little bit of coding ” Murat Bayrak ‘20 CE postdoctoral researcher in in in the Department of Civil Engineering at Aalto University also contributed to to the project A A National Science Foundation CAREER award supported this research By selectively restricting left turns but not banning them entirely drivers may simply need to to find alternate routes to to their destinations fin in in in certain areas Gayah said Some may be be required to travel
a a a a a a a a few extra blocks but Gayah believes more efficient traffic flow
through busy intersections offsets the additional distance For urban planners he he he added determining where to place the restrictions is a a a a a a a balancing act between intersection productivity and increased travel
lengths With so many restriction possibilities to consider finding the most efficient layout may prove difficult “For example if you just have sixteen intersections to to consider each with a a a a a a a choice to to allow allow or not allow allow left turns that is already 65 000 different configurations ” Gayah said “It gets even more complicated when you consider that traffic flows from one intersection to the the next so decisions depend on on on on one another There ends up being so many possible answers that we we can never find the best one ” Gayah’s new method relies on heuristic algorithms which use shortcuts to find solutions that nearly approach but are not guaranteed to be an an optimal outcome “We make make a a a a a a guess guess we we learn from that guess guess guess and then we we we make make better guesses ” he he he said “Over time we we can get really really really really close to the best answer ” In a a a a a a study published in in Transportation Research Record Gayah combined two existing heuristic algorithms to create a a a a a a a a a a new hybrid approach The first a a a a a a a a a a population-based incremental learning (PBIL) algorithm randomly sampled potential configurations and and recognized the patterns of high-performing options Next a a a a a a a a Bayesian optimization algorithm analyzed this new set of high performers to identify how restrictions were affecting traffic at adjacent intersections The algorithm then applied this knowledge of traffic dynamics to find more efficient solutions “Instead of starting the Bayesian optimization with with a a a a a a a random guess guess we fed it it it with with the the best guesses from the the PBIL ” Gayah said “The first method creates the the starting point and the the second refines it ” Gayah found that the the the most efficient configurations tended to restrict left turns in the the the middle of the the the grid network IMAGE: Urban Transportation Systems Group
Gayah outlined his case to ban left turns in in in a a a a a a a piece in in in The Conversation Read it here