Real-time eco-driving strategy for intelligent connected vehicles under mixed traffic environments

Eco-driving in mixed traffic environments faces challenges in real-time multi-objective optimization. This article proposes a hierarchical adaptive multi-intersection eco-approach and departure (AM-EAD) strategy. The upper layer generates a speed profile for multiple intersections by integrating information of intersections and preceding vehicles constraints. Especially, for car-following, a safe distance is determined using a variable time headway (VTH) model with a Kalman filter. The lower layer employs model predictive control (MPC) to coordinate speed tracking, car-following, and parking control via predictive horizon adaptation, achieving multi-objective optimization while improving computational efficiency. Validation is performed using a co-simulation platform integrated with real-world road data. Hundred randomized simulations show that compared to isolated intersection EAD (I-EAD) and rule-based EAD (R-EAD) benchmarks, AM-EAD reduces energy consumption (EC) by 8.39% and 19.55% and shortens travel time by 1.89% and 8.00%. Hardware-in-the-loop (HIL) experiments confirm real-time performance with an average computation time of 8.21 ms.