Off-campus SJSU users: To download campus access theses, please use the following link to log into our proxy server with your SJSU library user name and PIN.
Publication Date
Spring 2024
Degree Type
Thesis - Campus Access Only
Degree Name
Master of Science (MS)
Department
Computer Engineering
Advisor
Stas Tiomkin; Magdalini Eirinaki; Younghee Park
Abstract
Traffic congestion poses a persistent challenge, particularly with the rapid urbanization of metropolitan cities. Existing centralized solutions are limited in scalability, necessitating a shift towards decentralized approaches. This proof-of-concept study delves into leveraging intrinsic motivations, notably empowerment, to optimize autonomous car behavior, thereby alleviating traffic congestion and enhancing traffic flow. Within conventional traffic dynamics models, where spontaneous traffic jams emerge and extend over long distances, we explore the proposed method which operates without explicit coordination among the drives, functioning in a fully decentralized manner. Such decentralization is essential for any solution that ensures both drivers’ privacy and the viability of the traffic flow in the face of network communication failures or delays. Drawing from the Nagel-Schreckenberg traffic model, we substitute a portion of cars with empowerment-driven (intrinsically motivated) agent cars. Through extensive simulations, we demonstrate that our proposed method significantly enhances traffic flow (by ≈ 27%), and reduces average traffic jam time (by upto ≈ 95%), showcasing the effectiveness of our approach in addressing this critical urban challenge.
Recommended Citation
Papala, Himaja, "Decentralized Traffic Congestion Control Using Intrinsically Motivated AI Agents" (2024). Master's Theses. 5522.
DOI: https://doi.org/10.31979/etd.c7bd-qt2r
https://scholarworks.sjsu.edu/etd_theses/5522
