Semi-autonomous driving features are understood to have an impact on vehicle fuel efficiency, but current fuel economy testing methods don’t account for this technology. According to new research from Carnegie Mellon University’s (CMU) Department of Civil and Environmental Engineering, the way testing methods are developed and implemented could have a dramatic impact, in terms of creating the right incentives for automakers to develop autonomous features in ways that make cars more efficient.
CMU’s tests focused on early semi-autonomous features that are already being incorporated into new cars, which fall into Level 2 and Level 3 categories for automation. In some scenarios, these features created a fuel efficiency gain of up to 10 percent. In other scenarios, these features were neutral, and in some there was even a decrease in efficiency of 3 percent.
“We want the EPA to come up with a standard testing regime for features of early automation, so that manufacturers have an incentive to design them in a way that will boost efficiency.”
According to Constantine Samaras, an Assistant Professor of Civil and Environmental Engineering at Carnegie Mellon University who led the research, it’s not only important to understand how these technologies impact fuel efficiency, but now is the time to develop rules and implement them into published ratings. “The important thing is that we understand this could be a decent improvement, or even a potential degrading of fuel economy, depending on how we program the cars,” Samaras said. “We want the EPA to come up with a standard testing regime for features of early automation, so that manufacturers have an incentive to design them in a way that will boost efficiency.”
Humans are highly inefficient drivers, due to their tendency to drive non-optimal routes and brake or accelerate irrationally. CMU’s research observed the fuel economy impacts of semi-adaptive cruise control technology, as well as cruise control with some vehicle-to-vehicle communication. The paper proposes the addition of “Automated Drive Cycles” to the fuel economy testing regimen, and offers a testing methodology for regulators to consider.
“If there’s no standard testing procedure, the manufacturer won’t get credit for the fuel economy benefits of these features,” says Samaras. “If they don’t get credit, they can’t put it on the window sticker, and they can’t include in their corporate average fuel economy (CAFE) portfolio. Thus, they don’t have incentive to design autonomous features in a way that maximizes fuel economy. They might do it anyway, and it would make sense if they do, but they also might not—they might design the car to drive in a way that optimizes for speed, or mimics the acceleration of a human driver.”
When it comes to developing these rules, Samaras argues, sooner is better. “The rulemaking process is going to be lengthy, we’ve demonstrated that even with Level 2-3 technology. It makes a big difference, and now is the time to start a discussion around how to incorporate these features into fuel economy testing, so that manufacturers can get credit and design autonomous features for higher miles per gallon (mpg).”
Other observers of fuel economy regulations have expressed concern that existing autonomous driving technologies are still in their infancy, and it’s too early to incorporate them into existing fuel economy standards. “There’s a tradeoff in developing standards while the technology is still evolving,” says Samaras. “But I think outlining a strategy that encourages both innovation and fuel efficiency is possible. Without a way to measure fuel economy benefits of early automation, we risk having manufacturers design the first generation of automated vehicles without prioritizing fuel economy.”
According to the research, if regulators fail to align manufacturer incentives with improving fuel economy, automakers “are likely to make vehicle control decisions that increase vehicle desirability at the cost of fuel efficiency.” The paper also notes that the National Transportation Safety Board is considering if various partially-autonomous technologies should be included as standard vehicle features for safety reasons. This “would enhance the importance of understanding their impacts on vehicle fuel economy,” given their more widespread deployment.
Off-cycle technology credits
Manufacturers can petition for off-cycle technology credits to increase a vehicle’s CAFE fuel economy rating if they demonstrate current testing does not capture some of the efficiency gains provided by “new and innovative technologies.” The research points to three potential challenges to this approach. The first is the fact that these credits should only apply to non-standard technologies, and many semi-autonomous driving features are already too widespread to be eligible. The second is that the process will not create equivalent testing across manufacturers. Finally, manufacturers are not allowed to include the improvements on fuel economy ratings stickers that inform the consumer, or advertise the efficiency benefits of the technology.
In other words, the off-cycle technology credit program is not an appropriate mechanism to address the scope of autonomous vehicle technologies and fuel economy standards. “We can’t run autonomous vehicles through the off-cycle technology credit program. We need EPA to get ahead of this, and do so in a way that enables innovation and advancement of this technology,” says Samaras.
Using off-cycle technology credits to accommodate certain autonomous vehicle technology is not even permitted in the fuel economy rule for 2017-2025 model year vehicles
In fact, using off-cycle technology credits to accommodate certain autonomous vehicle technology is not even permitted in the fuel economy rule for 2017-2025 model year vehicles. The final rulemaking lists a number of rejected requests for off-cycle technology credits, among them, “congestion mitigation/crash avoidance systems.” It also says, “In the case of crash-avoidance technology, we are prohibiting off-cycle credits for these technologies under any circumstances.” CMU’s research focused on semi-adaptive cruise control technology, which does not fall into this category. However, there are some crash avoidance technologies which would have the potential to reduce overall fuel consumption by reducing the congestion caused by accidents.
Efficiency gains versus demand increases
When looking at a transition to widespread use of fully-autonomous vehicles, questions remain regarding the tradeoffs between efficiency gains and increased travel demand. Autonomous vehicles will increase demand from populations that currently don’t drive—the very old, the very young, the blind, certain other groups with disabilities—and is likely to boost travel overall by increasing the convenience of using a car. Estimates vary, but it’s anticipated that some or all of this demand increase will be offset by efficiency improvements. According to Morgan Stanley research, a “utopian” shift to autonomous vehicles in which they become the new norm would bring about $150 billion per year in fuel savings in the United States alone. That figure is on top of the fuel economy improvements already put into place through 2025. Morgan Stanley expects that removing the human inefficiency of driving could provide an additional 20-to-30 percent improvement, while fuel economy could be improved by an additional 50 percent thanks to lightweighting and improved aerodynamics.
Companies like Uber who deploy an autonomous vehicle service will place a large emphasis on the per-mile cost of driving, which is minimized through electric drive.
Others emphasize the fact that the vast majority of autonomous vehicles are expected to be electric, for a number of reasons. “Self-driving cars also allow electric vehicles to diffuse faster, because the increased efficiency enables you to get more out of a battery,” says Samaras. “There might be other benefits—with driving infrastructure, shared rides, etc., autonomous electric vehicles will change the landscape for the better in terms of transportation.” Additionally, early deployment of autonomous vehicles is expected to occur largely through ride-sharing services, in geographic regions which are primed for driverless cars. Companies like Uber who deploy an autonomous vehicle service will place a large emphasis on the per-mile cost of driving, which is minimized through electric drive, creating longer-term potential for driverless cars to reduce fuel consumption.