Predictive Eco-Driving Control for CAVs in a Traffic-in-the-Loop Environment

The transportation sector accounts for a substantial share of global energy consumption and carbon emissions, making the development of energy-efficient vehicle control strategies an essential step toward sustainable mobility. With the rapid advancement of Connected and Automated Vehicles (CAVs), predictive control technologies are increasingly applied to improve vehicle efficiency through the anticipation of future road and traffic conditions. In order to take full advantage of this feature, this study develops a predictive eco-driving control framework based on Model Predictive Control (MPC) within a traffic-in-the-loop simulation platform. The framework enables dynamic interaction among vehicles, surrounding traffic, and infrastructure, thereby optimizing speed and acceleration trajectories in real time. A revised Enhanced Driver Model (rEDM) is first established as a comparison baseline, followed by a basic MPC controller that optimizes longitudinal motion while maintaining safety and comfort. Compared with the rEDM, the MPC achieves an 8÷15 % reduction in wheel-end energy consumption, confirming its effectiveness in improving energy efficiency through predictive optimization. The MPC controller is then extended with three environment-aware functions. The Road Grade Preview (RGP) anticipates slope variations to smooth torque demand and provides an additional 1÷4 % energy saving. The Green Wave (GW) function integrates SPaT data to coordinate the vehicle with green window, reducing stop-and-go, and idling losses and achieving up to 12 % improvement in energy efficiency. The Road Curvature Preview (RCP) module constrains cornering speeds based on a 0.15 g lateral acceleration limit, enhancing safety and comfort during turns. These functions are ultimately combined into a multifunctional MPC (multi-MPC) that unifies grade, signal, and curvature awareness within a single controller framework. In the urban scenario, the multi-MPC achieves up to 51 % energy savings compared with the baseline rEDM while maintaining stable, compliant, and safe driving performance. When evaluated in a new mixed urban–rural route of 10.8 km, the controller continues to operate reliably, achieving approximately 27 % overall energy reduction with respect to rEDM. However, its benefit in rural conditions becomes less pronounced. This outcome highlights that, although the proposed MPC framework exhibits strong robustness and adaptability in complex urban environments, its energy-saving performance under diverse traffic conditions still has room for further improvement.