The Green Co-Driver: A Human-Interactive Self-Driving System That Improves the Energy Efficiency of Road Vehicles

This paper presents the design and experimental evaluation of a green driving function that balances energy efficiency and human driver expectations through a user-adjustable trade-off parameter. The function operates as a high-level advisory layer that provides eco-driving speed recommendations on top of an Advanced Driver Assistance System controlling longitudinal motion, making it portable across different advanced driver assistance and self-driving architectures. The system was tested in two different driving simulators with a total of 72 participants (aged 18-65), analyzing both quantitative performance metrics and subjective user evaluations. Quantitative metrics (e.g., energy consumption, travel time, and speed profiles) indicate that, in the considered simulator-based scenario and at the trade-off setting preferred by participants, increasing travel time by approximately 13.5% reduces energy consumption by about 34%, which corresponds to an increase in vehicle range of approximately 52% for the given vehicle characteristics. Additionally, the system exhibits negligible variability across repeated trials under identical simulated conditions, in contrast to the substantial variability observed in human coasting behavior, thereby eliminating human variations that negatively impact energy consumption. Subjective results show that, while manual driving is perceived as more fun, people are willing to adopt the system when confronted with a large extension of vehicle range at minimal travel time cost.