Autonomous cars -- vehicles in which a computer assumes control at least part of the time -- are not only coming, they're here.
In bits and pieces, automakers have been rolling the technology for years. It steps in to help cars brake quickly, avert collisions and even take the wheel for very short periods of times.
But that's very different from highways full of computer-operated vehicles moving in tandem, efficiently and nearly accident-free. What stands between today and that utopian future? We talked to two scientists who can imagine how all the moving parts might fit together.
Q: What might partial integration of fully autonomous vehicles look like? How do we mix human drivers and their computer counterparts?
Lora Weiss, Ph.D.
Lab Chief Scientist and Technical Director for Autonomous Systems
A: There are all these terms being thrown around, like driverless car vs. autonomous car vs. robotic car vs. self-driving. First of all, it’s going to be a very long time before there will be driverless cars. I believe they will be more self-driving.
We are definitely moving in that direction as a society; for example, we started a long time ago with cruise control and automatic braking. Currently, cars have lane change warning systems and soon they will be able to do lane changing as well. There are cars out there today that, at slow speeds, if they sense an impending accident, will begin to brake before the driver senses the incident. Therefore, we are certainly moving in the direction of much more automation and aspects of self-driving in cars that are coming out today.
Where you will start seeing some more advances is where the vehicle interacts with the infrastructure, called V2I (Vehicle to Infrastructure). This is where the vehicle talks to traffic lights and other roadway infrastructure. Right now, there is a challenge of vehicles talking to other vehicles, called V2V (Vehicle to Vehicle) communication, because not every vehicle is equipped with the communication information. But as vehicles begin to speak with the infrastructure (roadways, street signs and emergency systems) there will be a lot of improvements in safety especially.
Umit Ozguner, Ph.D.
Professor of Electrical and Computer Engineering, TRC Chair in Intelligent Transportation Systems, Control and Intelligent Transportation Research Group
A: Partial integration of partially autonomous vehicles (with Advanced Cruise Control and Active Lane Departure Warning) already exists. “Stand alone” fully autonomous vehicles in human-driven traffic won’t be that much different. It will be like more Google Cars.
Or there will be some segregated lanes, especially for trucks, but possibly also for autonomous cars that have the required “autonomy options”.
Another possibility is using autonomous vehicles in “closed” environments: airports, parks, campuses, retirement villages.
The most complicated possibility is a mixture of “collaborating self-driving cars” within the human traffic. These could be relying on information and sensing provided by external sources and make decisions based on that. (Think of a car broadcasting “I think my driver will run this red light!”. How would you react?)
Q: It's 2030: I live in a suburb, own a fully autonomous car and have experienced a car accident. What do you envision the structure of the car insurance coverage to be? Who will be the responsible financial party?
Umit Ozguner, Ph.D. - Professor of Electrical and Computer Engineering at Ohio State
A: This will be complicated. It will depend on whether the accident was due to driver fault (use in an ineligible area, sudden intervention, ignoring maintenance, etc.), or to design flaws (cases/events that slipped designers’ imagination and the requisite certification); or to operational errors (many possibilities based on who was supposed to tell whom, when and the effect of various resources being “down”).
Q: We conducted a survey of consumers and found that 20% were ready to give up driving once self-driving cars were established, but the remainder would rather be in control of their own cars.
Lora Weiss, Ph.D. - Technical Director for Autonomous Systems at Georgia Tech
A: Ford had commissioned a study about a year go where nine out of 10 drivers wanted to assist technology and not have a self-driving car. This might just be due to a sense of familiarity. How quickly did people adapt to cruise control and anti-lock brakes? But today, it’s common. Fifty percent of people have either fallen asleep behind the wheel or know someone who has fallen asleep; so there are tremendous safety benefits of implementing assisted driving technology and it comes down to people becoming familiar and comfortable with the technology.