Machine Learning for Individualized Vehicle Eco-Routing
Barron Associates Inc. (BAI) and its subcontractor, Southwest Research Institute (SwRI), are using advanced machine learning (ML) approaches to generalize and enhance aspects of the SwRI Connected Powertrain Group’s NEXTCAR (NEXT generation energy technologies for Connected and Automated on-Road vehicles) program, which is funded by the Department of Energy. The technology resulting from this research effort will result in NEXTCAR-ML, which will set the stage for widespread adoption of vehicle eco-routing technology and realize the commensurate economic and environmental benefits. Development of highly-accurate, a priori fuel consumption models for every possible vehicle over every possible drive cycle is financially and logistically impossible. Therefore, the research team is using ML techniques to synthesize individualized predictions of energy consumption using on-vehicle data collected during real-world driving experience. These ML-based models will evolve with learning over time to accurately predict route-based energy consumption. ML-based models will intrinsically capture vehicle loading, powertrain condition, route high-frequency speed content, and human (or, in the future, autonomous) driver characteristics. The envisioned Phase I and Phase II research effort will include computer simulation, dynamometer testing, track testing, and real-world route testing. The resulting NEXTCAR-ML technology will help capture the vast potential energy savings that is possible using widespread eco-routing. SwRI’s NEXTCAR research produced highly-encouraging results regarding the efficacy of eco-routing. Eco-routing operates at the macroscopic level to identify the most energy efficient route to a destination based on the specific vehicle and powertrain. Using coarse information about traffic congestion, grade, stop signs, school zones, charging opportunities, etc., it may be found that the shortest or fastest route is not the most energy efficient. Using dynamometer testing, SwRI found instances of 20% energy savings using eco-routing, which was much higher savings that originally anticipated. In many cases, the travel time penalty was minimal. Consumer adoption of eco-routing technology and driver selection of the most energy-efficient route are expected to occur at high rates. The overall goal of this research is to use machine learning to generalize NEXTCAR eco-routing approach to provide energy savings for any on-road vehicle even without a detailed a priori vehicle and powertrain model. The generalized technology leverages machine learning algorithms to individualize eco-routing options based on actual vehicle and driver experience. The highly successful Phase I effort has unequivocally demonstrated the technical feasibility of the proposed NEXTCAR-ML technology.
- Research Organization:
- Barron Associates, Inc.
- Sponsoring Organization:
- USDOE Office of Science (SC)
- DOE Contract Number:
- SC0020891
- OSTI ID:
- 1805224
- Type / Phase:
- SBIR (Phase I)
- Report Number(s):
- DOE-BARRON-20891
- Country of Publication:
- United States
- Language:
- English
Similar Records
RouteE: A Vehicle Energy Consumption Prediction Engine
RouteE: A Vehicle Energy Consumption Prediction Engine