Predictive Data-Driven Automotive Control

Default ARPA-E Project Image

Project Term:
03/03/2017 - 03/21/2024

Critical Need:

Modern drivers are skilled at anticipating and reacting to the behavior of nearby vehicles and the environment in order to travel safely. Nevertheless, all drivers operate with an information gap – a level of uncertainty that limits vehicle energy efficiency. For instance, safe driving demands that drivers leave appropriate space between vehicles and cautiously approach intersections, because one can never fully know the intentions of nearby vehicles or yet unseen traffic conditions. Closing this information gap can enable vehicles to operate in more energy efficient ways. The increased development of connected and automated vehicle systems, currently used mostly for safety and driver convenience, presents new opportunities to improve the energy efficiency of individual vehicles. Onboard sensing and external connectivity using Vehicle-to-Vehicle (V2V), Vehicle-to-Infrastructure (V2I), and Vehicle-to-Everything (V2X) technologies will allow a vehicle to “know” its future operating environment with some degree of certainty, greatly narrowing previous information gaps. By providing the ability to predict driving conditions, these technologies could operate the vehicle powertrain (including the engine, transmission, and other components) more intelligently, generating significant vehicle energy savings.

Project Innovation + Advantages:

The University of California, Berkeley (UC Berkeley) will lead a team that includes Sensys Networks and Hyundai America Technical Center to develop a novel control technology to reduce energy consumption of a plug-in hybrid electric vehicle by at least 20% without changing its drivability. Through connectivity with other vehicles and the roadway, the vehicle will access data such as signal phase and timing, traffic queue length and position, traffic volume and speed, and position and speed of nearby cars. The powertrain and vehicle dynamic controllers developed in this project will utilize this data to optimize how the plug-in hybrid test vehicles operate by adjusting parameters such as vehicle speed, electric motor torque, and battery charging power. The technology will be demonstrated with a fleet of vehicles in three applications: cooperative adaptive cruise control, speed harmonization with merging vehicles, and optimal approach/departure at intersections with traffic signals. The ability to work in real-time with a large number of factors and scenarios is enabled by computation conducted both onboard the vehicle and off-board using cloud-based computers. The team combines expertise in algorithm development and predictive controls from UC Berkeley with a leading technology for roadway sensing and vehicle to infrastructure (V2I) communication from Sensys Networks. Hyundai, a major global car manufacturer, will provide a state-of-the-art plug-in hybrid vehicle platform, extensive vehicle testing capability, and also a path to commercialization for the proposed controller technology into the high-volume light-duty vehicle market.

Potential Impact:

If successful, UC Berkeley’s project will enable at least an additional 20% reduction in energy consumption of future connected and automated vehicles.


These innovations could lead to a dramatically more efficient domestic vehicle fleet, lessening U.S. dependence on imported oil.


Greater efficiency in transportation can help reduce sector emissions, helping improve urban air quality and decreasing the sector’s carbon footprint.


Innovations would further solidify the United States’ status as a global leader in connected and automated vehicle technology, while a more efficient vehicle fleet would reduce energy cost per mile driven and bolster economic competitiveness.


ARPA-E Program Director:
Dr. Marina Sofos
Project Contact:
Prof. Francesco Borrelli
Press and General Inquiries Email:
Project Contact Email:


Hyundai America Technical Center
Sensys Networks

Related Projects

Release Date: