The paper introduces a bi-level control strategy for managing lane changes in weaving sections using connected and automated vehicles (CAVs). The upper level employs deep reinforcement learning to determine control weights, while the lower level uses model predictive control within each CAV. The proposed method outperforms existing benchmarks in a case study inspired by a real weaving section in Basel, Switzerland.
Til et andet sprog
fra kildeindhold
arxiv.org
Vigtigste indsigter udtrukket fra
by Longhao Yan,... kl. arxiv.org 03-26-2024
https://arxiv.org/pdf/2403.16225.pdfDybere Forespørgsler