Validation of an Energy-efficient longitudinal vehicle motion control based on model predictive control
Automated driving functions open up a high potential for energy savings, especially in dynamic urban traffic, through optimized, automated vehicle longitudinal guidance. The complex interaction with a stochastic, dynamic traffic environment places high demands on the control algorithms to be developed. Model Predictive Control (MPR) has emerged as a promising approach for implementing optimized longitudinal guidance. To generate energy-efficient, safe and comfortable trajectories for longitudinal guidance, the vehicle dynamics as well as the predicted motion profiles of the road users are explicitly included in an optimization. In previous research, a simulation study showed that a driving strategy based on MPR could save up to 44% of drive energy compared to a human driver, depending on the drive type and traffic situation.
The transfer of simulation-based approaches into a real validation is challenging. In order to enable the transfer from research to real application, simulatively developed driving functions are to be validated and verified in real driving scenarios within the scope of this project. For this purpose, a production vehicle equipped with basic environment sensor technology will be extended by a Rapid Control Prototyping (RCP) control unit on which a developed MPR strategy will be implemented. Based on measurement data from specified test scenarios, the model for state estimation of dynamic driving scenarios can be adaptively adjusted and extended.
The long-term goal is to use the extension of environmental perception to develop interactive trajectory planning that takes into account a combination of the reaction of road users and the driver's own planning for a long time horizon. In addition to safety and comfort aspects, this will further improve energy efficiency.