Energy and flow effects of optimal automated driving in mixed traffic: Vehicle-in-the-loop experimental results

Ard, Tyler; Guo, Longxiang; Dollar, Robert Austin; Fayazi, Alireza; Goulet, Nathan; Jia, Yunyi; Ayalew, Beshah; Vahidi, Ardalan · 2021 · Crossref

DOI: 10.1016/j.trc.2021.103168

archive: archived pipeline: cataloged verified

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

Summary

This paper addresses the challenge of improving energy efficiency and traffic flow in mixed traffic environments containing both Connected and Automated Vehicles (CAVs) and human-driven vehicles. While previous research demonstrated that automated vehicles could enhance network throughput and safety, there was a need to experimentally quantify energy savings in realistic, mixed-traffic scenarios without compromising safety or following headways. The authors propose an anticipative car-following algorithm based on Model Predictive Control (MPC) that leverages connectivity when available and probabilistic constraints when interacting with unconnected human drivers. To evaluate this approach, the researchers developed a Vehicle-in-the-Loop (VIL) testing environment. This framework embeds physical vehicles—a Mazda CX-7 with a combustion engine and a Nissan Leaf electric vehicle—into a virtual traffic microsimulation using PTV VISSIM. The VIL architecture consists of a server layer running the simulation, a client layer handling high-level trajectory planning and low-level execution, and a hardware layer with pedal and steering actuators. The physical vehicle’s position and velocity are synchronized with the virtual environment in real-time via RTK-GPS and UDP communication. The high-level controller uses MPC to minimize acceleration and headway tracking errors. In the connected case (MPC-C), the ego vehicle receives the planned trajectory of the preceding CAV. In the unconnected case (MPC-U), the system predicts the preceding human driver’s motion and employs chance constraints to manage uncertainty, balancing safety with traffic compactness. Low-level control combines classical feedback with data-driven nonlinear feedforward maps to track acceleration commands. Baseline comparisons were made against the Wiedemann 99 and Intelligent Driver Model (IDM) controllers, tuned to replicate human driving behavior. The experimental results demonstrate that the proposed MPC-based automated driving strategy significantly improves energy economy. Across various scenarios, including city and highway drive cycles and emergent highway traffic, the automated controllers achieved up to 30% improved energy efficiency compared to the human-driver baselines. These energy savings were realized for both the internal combustion engine and electric vehicles. Crucially, these improvements were achieved without sacrificing safety or increasing the following headway, indicating that the probabilistic constraints effectively managed the uncertainty of human-driven preceding vehicles while maintaining traffic flow. The significance of this work lies in its experimental validation of anticipative control strategies in a realistic mixed-traffic setting. By demonstrating substantial energy reductions without compromising safety or throughput, the study supports the integration of connectivity and predictive control in automated vehicles. The VIL framework provides a scalable method for evaluating cyber-physical traffic systems, offering insights into how CAVs can optimize energy use and traffic flow even when interacting with non-connected human drivers. This contributes to the broader goal of deploying automated vehicles that enhance both environmental sustainability and transportation network efficiency.

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.

Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.

Topics

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

• situational awareness

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).

• Methodological Resource: tool software
• Theoretical Contribution: computational model