Testing Road Grade-Aware Benefits of Koopman Model Predictive Control for Energy-Efficient Driving

This study presents the implementation and evaluation of a road-grade-aware control strategy for energy-efficient driving on the Cadillac Lyriq electric vehicle (EV) as a test platform. Leveraging a Koopman Model Predictive Control (KMPC) framework, the control approach developed integrates real-time road grade variations into its predictive optimization by using Legendre polynomial-based modeling. A dynamometer-based vehicle-in-the-loop (VIL) testing setup was used to rigorously assess the efficacy of the control system across a range of scenarios characterized by varying road grade profiles and target cruising speeds. Comparative analyses with baseline controls show significant energy savings, with KMPC effectively using downhill momentum for kinetic energy storage to enhance uphill driving efficiency. This paper provides comprehensive details of the control system design, its dynamometer-based validation, and the quantified energy and time metrics to demonstrate the practical viability of the grade-aware strategy in improving EV driving efficiency.