Vehicle-Traffic Control with Limited-Capacity Connected/Automated Vehicles

Ban, Xuegang (Jeff); Guo, Qiangqiang; Angah, Ohay; Liu, Zhijun · 2020 · ROSA P / Connected Cities for Smart Mobility toward Accessible and Resilient Transportation Center (C2SMART)

archive: archived pipeline: cataloged verified

Get this paper ↗ (full text — opens at the source; we link to it, we don't host it)

Summary

This research addresses the challenge of controlling vehicle and traffic systems in environments with limited penetration of low-level connected and autonomous vehicles (LCAVs). While prior studies focused on fully automated vehicles with high connectivity, this work targets the near-term reality of Level 2 or 3 automation. The study aims to develop and test integrative vehicle-traffic control methods that improve fuel efficiency and reduce traffic delays, moving beyond traditional traffic simulation to a more realistic testing environment. The investigation comprises three primary components. First, the authors developed a Hybrid Deep Q-learning and Policy Gradient (HDQPG) algorithm for Eco-Driving. This reinforcement learning approach controls both longitudinal operations (continuous acceleration/deceleration) and lateral decisions (discrete lane changes) for a single LCAV on signalized corridors. To ensure safety and decision consistency, the authors implemented a “checking-feedback-learning” (CFL) framework. Second, a Vehicle-in-the-Loop (VIL) simulation platform was constructed to reduce real-world testing costs. This platform integrates four components: SUMO for microscopic traffic simulation, Unity for vehicle dynamics and visualization, an AWS DeepRacer car to execute control commands, and a control center for coordination. Third, a dynamic Highway Capacity Manual (HCM) method was designed to adjust signal timings based on estimated CAV penetration and traffic volume. This signal control method was combined with the HDQPG Eco-Driving algorithm to create an integrated vehicle-traffic control system. Numerical experiments demonstrated that the HDQPG algorithm outperformed existing RL-based Eco-Driving methods, successfully learning fuel-saving strategies comparable to model-based approaches while handling unusual driving conditions. Testing the integrated vehicle-traffic control method on the VIL platform yielded significant improvements: the controlled vehicle experienced reduced fuel consumption, and all vehicles on the corridor experienced reduced delays compared to baseline scenarios. The VIL platform proved effective in validating these algorithms in a semi-realistic environment that accounts for vehicle dynamics and sensor data. The study concludes that integrative control methods combining adaptive signal timing and reinforcement learning-based Eco-Driving can effectively enhance traffic efficiency and fuel economy even with limited CAV penetration. The successful deployment of the VIL platform provides a viable pathway for testing such algorithms before real-world implementation. The authors recommend future research expand the Eco-Driving algorithm to multi-agent systems capable of controlling multiple CAVs simultaneously and developing fully integrative frameworks that allow direct communication and coordination between vehicle control and traffic signal systems.

Key finding

The integrated vehicle-traffic control method reduced fuel consumption of the controlled vehicle and decreased delays for all vehicles in the corridor.

Methodology

simulator

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via bulk_ingest_rosap on 2026-05-23 (6 acquisition events logged).

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• traffic density

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Theoretical Contribution: computational model