United States Department of Transportation
ROSA P serves as an archival repository of USDOT-published products including scientific findings, journal articles, guidelines, recommendations, or other information authored or co-authored by USDOT or funded partners.
As a repository, ROSA P retains documents in their original published format to ensure public access to scientific information.
i
Drivers’ Attitudes Toward Rerouting: Impacts on Network Congestion
-
2023-08-01
[PDF-551.18 KB]
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Research Report (2021 – 2023)
-
Corporate Publisher:
-
Abstract:Bifurcation, a phenomenon that always occurs in the transition phase between free flow and congestion in the macroscopic fundamental diagram (MFD), have been suspected of decreasing network efficiency. Some studies have shown that rerouting behavior will effectively postpone the occurrence of bifurcation. On the other hand, deep neural networks within reinforcement learning algorithms have been used in traffic control in recent years and produced important breakthroughs. However, recent work has found that it is difficult to learn something in extremely congested networks due to the congested network property. This report investigates how rerouting behavior affects the bifurcation phenomenon by using deep reinforcement learning (DRL) in drivers' behavior guidelines. We show that i) the learning efficiency of DRL is affected heavily by the congestion level, and ii) the result of the convergence process indicates that DRL didn't take good effect at both low- and high-density levels. These findings lead to a contradiction. We used two different models related to drivers’ rerouting tendency to ascertain the key factors in the training process. The study demonstrated the weakness of DRL on drivers' behavior control, and we hope it will be a useful reference to better understand the effect of DRL on transportation and to advance research in this area.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
