Last year, the number of motor vehicle fatalities topped 40,000 in the United States, with most involving some form of human driver error. Professor Daniel McGehee of the University of Iowa College of Engineering and his team are working to lessen the severity of these crashes–if not prevent them altogether–through the use of automated driving technologies. Many of these promising technologies are beginning to come to fruition, raising new policy issues in the process. Questions remain about how the law will respond to these developments in the automotive industry. It’s time to bring new lawyers up to date in this space.
On Friday, September 21, Prof. McGehee visited the University of Iowa College of Law to discuss these questions, bringing together law students and professors to learn about the past, present, and future of automated driving systems.
Vehicle Automation Classifications
Prof. McGehee introduced the subject by describing the six different levels of automation that the National Highway Traffic Safety Administration (NHTSA) uses to classify vehicles:
- Level Zero – No Automation – The vehicle has no automation capabilities.
- Level One – Driver Assistance – There is minimal, occasional automation of the vehicle (for instance, adaptive cruise control).
- Level Two – Partial Automation – The vehicle includes technology to assist the driver. The vehicle might put on the brakes automatically, or, if a driver drifts to another lane, it will gently steer the vehicle back into its lane.
- Level Three – Conditional Automation – In some driving modes, the vehicle is automated. The driver can push a button and the car drives itself; however, the driver can monitor conditions and resume driving the vehicle if necessary.
- Level Four – High Automation – The vehicle can drive itself. In this mode, if the driver were to become disabled (for instance, by a heart attack), the car could get out of traffic, find safe harbor and park on its own.
- Level Five – Full Automation – The vehicle is fully automated, and the driver has no ability to take over (there is no steering wheel and no brake pedal).
Vehicle Automation: Past, Present, and Down the Road
While high profile tests and demonstrations have recently brought automated driving technologies into the public eye, the first “Level One” vehicle came on the scene 60 years ago: the 1958 Chrysler Imperial debuted as the first vehicle with cruise control. In 1998, “Adaptive Cruise Control” (ACC) became available for high-end cars using a laser radar system to adapt the speed of the driver’s car to that of the vehicle in front of them.
Today, vehicles through “Level Two” are in production. In 2017, lower cost vehicles began including advanced driver’s assistance systems that can “see” 360 degrees around the vehicle. Cruise control has also improved. What we now call “stop and go” ACC can decelerate all the way to a stop on its own. Techniques may vary by manufacturer, but the principle tends to be the same: a vehicle uses cameras to “see” foci like pedestrians, bicyclists, and other vehicles, on the road–and draws a box around each one. The computer measures the size of the boxes over time. If a box quickly grows bigger (gets closer) the computer will alert the driver or automatically puts on the brakes. In addition to front sensors, some vehicles have technologies that can sense a car coming from the side when a driver is turning for cross-traffic collision avoidance.
Ride-sharing industries have been in the news recently for testing automated vehicles, but these vehicles are generally tested in highly controlled areas. Fully automated vehicles that seamlessly pick up passengers at their homes are still decades away. Still, other technologies are much closer. All new Toyotas and Subarus are expected to have “stop and go” ACC technology by next year. In the next two to three years, all vehicles are expected to include this technology.
Shifting Gears to New Liabilities
The shift from manual to automatic driving means a shift in liabilities. Because of the limited–or lack of–automation in Level Zero, Level One, and Level Two vehicles, the driver is completely in control and in charge of the vehicle. Level Five is fully automated and essentially drives without an operator, so the driver is not in control and the manufacturer is responsible.
Liabilities for Level Four and Level Three vehicles are not as clear. A Level Four vehicle is responsible for itself when it is in the automated driving mode, and the CEO of one manufacturer has already said the manufacturer will be 100% responsible for their car if it crashes when it is in the automated driving mode. Because the driver of a Level Three vehicle remains available to monitor and take over the driving of the vehicle, there are ambiguities regarding liabilities.
One such ambiguity is the length of time necessary for a driver to take over a Level Three vehicle. This is one of many research topics Prof. McGehee is looking at with his team. Prof. McGehee’s team uses real-time state detection software inside of vehicles to determine how people sit, what their typical posture is like, and what they’re doing when they drive. Through self-learning algorithms, his team can understand what’s normal, and what’s not normal, so that it’s possible to know whether the driver is ready to take over the vehicle if needed. For instance, if a person is sitting on the driver’s seat with their legs folded underneath them, they are not immediately ready to take over the vehicle. The next question is: how long does it take for them to get their legs unfolded?
“To Detect and to Swerve”
Prof. McGehee and his team at the University of Iowa National Advanced Driving Simulator study more than driver body language and timing. As indicated by the words “To Detect and to Swerve,” adorning one of their test vehicles, road safety drives Prof. McGehee’s automated vehicle research. Other research by Prof. McGehee and his team includes driving through a blizzard with a robot at the wheel, aggressiveness algorithms to sense the aggressive nature of other drivers, research involving cannabis and driving, and a large body of research and development work and independent evaluation. As the largest public academic driving simulator in the world, industry academics come from around the world to research at National Advanced Driving Simulator.
In addition to a collection of 20 years of advanced driver’s assistance system technology, the National Advanced Driving Simulator benefits from a close relationship with the National Transportation Safety Board. The NTSB investigates 100% of automated vehicle crashes and shares this data with the researchers, who collect and analyze the data.
The National Advanced Driving Simulator is also home to a fleet of specialty vehicles, ranging from the first Volvo XC90 in North America (the first vehicle to be built from the ground up as an automated vehicle) to the advanced Lincoln MKZ. Their Tesla–specially instrumented for automated research–just completed its very first study of commuters, when it was loaned for a week at a time to people who commute between Cedar Rapids and Iowa City.
One Last Thought for the Road
Law students left Professor McGehee’s talk with a newfound understanding of the technologies and current legal questions that will give them a greater understanding of how of automated driving laws and policies may develop. As society speeds toward a future of automated driving systems, it is the law students of today who will help us get there safely.
Written by Kara D. Schwee