We're five minutes from starting. If you have a question send it now and I'll add it to the list. You can always add your questions during the chat.
All right, let's begin. I'm a web editor with The Globe and I'll be moderating the chat. Feel free to send your questions and comments to the experts; they'll have the floor for most of the discussion.
I'd like to welcome Professor Raj Rajkumar and Siri Agrell to the chat. Thanks for joining us.
Good to join the discussion.
Hi, thanks for hosting Stuart and for joining us Prof. Rajkumar
Here's the first question:
There's been some confusion as to how driverless cars work. Do they rely on a network of information with several cars communicating with each other? Or do they use a collection of information from devices like cameras and sensors mounted on the car?
I'm going to let Raj handle the technical questions!
Both - each driverless car will have sensors (lasers, radars, cameras, etc.) and drive-by-wire capabilities along with networking capabilities. Cars will have WiFi-like technology (in addition to 3G/4G networking capabilities).. The technology is called DSRC (Dedicated Short-Range Communications) - it is essentially WiFi based but operates at a slightly different point in the wireless spectrum to avoid conflicts with WiFi. Both safety and traffic applications are plentiful.
Sounds like the technology is well-established. How long until we see these cars on the road or in some kind of mass production?
Well, they've already been on the road in California. And some Chinese researchers have apparently tested their prototypes as well.
Completely driverless cars are at least 10-15 years away. However, many features will be percolating to high-end cars over the next few years (like driverless transportation on highways alone).
Here's a related question from Curtiss Law:
Would they be rolled out individually for those who choose to use them? Or would this be something that had to be uniformly introduced by all major car companies? Could a non-computer controlled car still be allowed on the road with say 75% computer controlled cars?
I was referring to production cars.
Individual carmakers will roll out products. Most accidents on the roads happen due to human error. Over time, as more cars become driverless and demonstrate their reliability, one can imagine that humans will be deemed to be safer *not* driving.
I don't think anyone's going to mandate that everyone must buy one!
I was with readers a bit before you two arrived. Questions from Mike Hunt and SimonC revolved around the social aspect. How can driverless cars overcome the fact that people just like to drive?
Yes, lots of people enjoy driving (including me!). But not all the time you are in a car. There are times when I have to drive long distances and I wish somebody else were driving, so I catch up on some sleep and/or work. It would take several years before all cars on the roads become driverless. So, there will be mixed driving for a while. Over the longer term, think of what happened to those who love to ride horses in a horse-and-carriage transportation system. Will there be "car ranches"?
Does every one like to drive? We had a piece in the paper today about the insane commute times Canadian drivers face. I wouldn' t mind being able to read the paper instead of screaming at my fellow motorists!
Here's a question about Google from JoelG.
A reminder to readers that we're accepting questions. Just type a display name into the comment box and submit your question. When we're ready, we'll post it for our experts.
Not to take credit away from Google but it is worth noting that the key players in Google's team are from Carnegie Mellon University (my employer) and Stanford University, both of which work closely with current automakers.
With regards to indemnification, every new technology goes through a transition period when the variables are still being understood, but in time, since overall safety is enhanced, the returns will be very hard to ignore.
SimonC asked a similar question earlier about the ramifications for accidents. Can we anticipate how the insurance industry will adapt? Would humans be liable for accidents caused by computer error? Would penalties go up for drivers who cause a crash after taking control from their computer system?
These seem like questions that need answers before we can think seriously about introducing computer-controlled cars.
One thing Raj pointed out to me when we spoke for the article, is that we already rely on many robotic controls, like cruise control and auto pilot when we fly. And people seem comfortable with those...
Obviously liability is a large issue and one that needs to be worked out. But I think it's funny that people are so focused on it when the idea is to bring the number of accidents down.
Some high-end cars TODAY will brake on their own if it senses that an accident is inevitable. It finally becomes a game of statistics (favoring driverless cars), appropriate monitoring mechanisms (like video "black boxes") and perhaps some rules by governing bodies. Insurance companies would like to see accidents go down too.
Here's a question from hi2pi about the social aspect.
Yes, I can imagine that this technology will be adopted first in (say) SIngapore or Qatar, small countries where things are perhaps easier to mandate. But once people taste the amount of time saved and the lower likelihood of accidents, the demand will make the supply happen too.
I think we're adapting so quickly to technological influences in our life it's hard to say how big a shift it will be. I don't think it will be a matter of one day these cars being available for sale, it will be a gradual shift toward a more automated form of driving.
Here's a question about the environment from Kaleigh.
Cars can become lighter (since they are safer) and smaller (one reason people like SUVs because they are perceived to be safer in a collision). Brings up mileage. WIth traffic info and coordination, less time is spent on the roads. Less pollution results.
The environmental benefit was one aspect that was raised by everyone I spoke to while researching this piece. I find the SUV point really interesting: that people wouldn't feel they need big gas guzzlers if they felt safer in their cars.
Here's a question about the implications of driverless cars from David.
These are two separate questions. One can imagine driverless vehicle lanes but they can use the other lanes as well (safely). There is also a lot of money to be made in building and maintaining these new generations of vehicles.
The other flip side to that is the lost productivity we already experience because of our commute times. The cost of congestion, accidents + traffic is already taking a profound toll on our economies.
I do! Raj, you've participated in the government led competition to develop a driverless car. Could you tell us a bit about that event -- who goes, etc. And also, are there specific tasks that seem to trip up driverless cars? Parallel parking? Drive throughs?
Great question. The last competition that our team from Carnegie Mellon won was in 2007. THe Pentagon agency named DARPA which ran the competition declared victory in the sense that they had shown that the 'impossible' is possible. No competitions have been held since. That competition did not include pedestrians, night-time driving, driving under bad road and weather conditions etc. Also, in the real world, things including computers, sensors and drive-by-wire actuators can fail. We are studying these at Carnegie Mellon.