Autonomous vehicle to be tested on area roads part of a global revolution
ONE DAY, TIME SPENT in the car will be spent catching up on our favourite TV shows, reading a book or even napping. That day isn’t today, but we’re getting there. When it arrives, it will bring with it major societal changes.
Technological development is the key, with a global research effort underway. Another step on the road to that future – good, bad or indifferent, depending on your take – will be visible close at hand, as the Waterloo Centre for Automotive Research (WatCAR) at the University of Waterloo rolls out on-road testing of its autonomous vehicle, a Lincoln MKZ hybrid sedan nicknamed Autonomoose.
The university got the go-ahead from the province last fall to test the technology on public roads. Now, researchers are ready to do just that.
It’s an exciting time for Krzysztof Czarnecki, a lead researcher and professor in Waterloo’s Department of Electrical and Computer Engineering, and his team at the Waterloo Intelligent Systems Engineering Lab (WISE Lab).
“It’s going to be a huge milestone for our project,” he says of the on-road tests set for mid-October.
His own area of research centers on decision making systems – for instance, the ability of a self-driving car to identify and predict pedestrian behaviour. There’s the hardware involved – the cameras and sensors – and then the software needed to interpret the data and act on it. That kind of artificial intelligence is the most difficult part of developing self-driving cars that can interpret an entire scene out on the road and react accordingly.
Where human drivers can interact with pedestrians by eye contact and body language, an autonomous vehicle will have to learn to deal with unpredictable behaviour, he notes.
Safety is a key factor, calling for robust and redundant systems with built-in failsafes.
Advances will be incremental, just as they’ve been with the introduction of technology into our cars thus far.
Engineers rate autonomous vehicles on a scale of 0-5, 0 being the traditional cars that required the driver for all inputs. We’ve seen levels 1 and 2, which range from power-assisted functions through to cruise control and lane-centering technology. The autonomous label doesn’t really apply until level 3, which is the next step now.
At level 3, the vehicle becomes capable of critical functions such as navigation, accelerating and braking, but all with close monitoring by the driver – no napping just yet.
Czarnecki notes there are no level 3 cars in production, but makers such as Tesla and Audi say they’ll be offering up such vehicles in the next year or so.
“That’s where you can take your hands off the steering wheel, but you have to pay attention,” he says. “You have to be ready to take over at any time.
“We might be able to see level 3 perhaps as soon as next year.”
Level 4 sees cars become generally autonomous – reading a book or even a nap are possible, but under limited circumstance, such as typical commutes.
For all kinds of roads – say the dirt and gravel roads of the rural areas – and conditions, we’ll have to wait for Level 5.
Such advances are five to 10 years out, Czarnecki predicts.
For now, researchers are focusing on getting the technology to the point where it can deal with the multiple challenges of changing conditions. In rolling out the Autonomoose next month, UW teams will start on quiet roads before ramping up to busier streets and winter conditions, making it more difficult for the car.
“The main challenges are really there: winter driving, night-time driving – making sure the car does the right thing in all conditions.”
The hardware and software challenges, particularly the artificial intelligence, are so complex that small steps, tweaks along the way, are the norm.
Czarnecki points to Google’s tests of autonomous vehicles, where disengagement episodes – when the human driver is asked to take over – occur about once every 5,000 miles, up from about every 2,500 miles a year ago.
Unlike human drivers, who all have to learn the skill individually when they come of age, autonomous vehicles can actually “learn” from the vast amounts of data gathered by researchers around the world. That allows for advances on many fronts.
While the safety of the technology is a factor in both its clearance for general road use and its acceptance by the public, driver safety is also a major, well, driver of autonomous vehicles: developed as expected, such cars would lead to fewer accidents.
“In my mind, there’s no doubt that a robotic vehicle will be safer,” says Czarnecki, noting computers don’t get tired or angry or inattentive.
That’s a big deal, as some 95 per cent of accidents are due to human error – “Many of those can be prevented.”
Getting to that point, however, will come with impacts well beyond safer, nap-filled commutes.
Autonomous vehicles have the potential to eliminate all driving jobs, from cabbies to transit workers. Let’s look at the trucking industry, one of the largest employers and a place where there are jobs for those without higher education. According to the American Trucker Association, there are 3.5 million professional truck drivers in the U.S. – it’s the most common job in more than half the states. Add to that another 5.2 million in jobs within the trucking industry. On top of that, there are millions more in related jobs, from logistics right on through to the diners and motels that cater to drivers.
Now, imagine all of the jobs disappearing as a result of driverless trucks.
The technology already exists today, just as it does
with driverless cars, another use of technology destined to displace jobs such as cabbies and couriers. Driverless buses and trains will eliminate the need for transit workers, many of them an increasing burden on governments and taxpayers.
Many industries are likely to undergo massive changes, as the technology is likely to be electric, eliminating much of the petroleum-based business for vehicles. Moreover, we may not even own individual cars, instead adopting a system based on a smartphone app summon- ing a vehicle as needed. (Our cars currently spend about 90 per cent of their time parked, unused). That would dramatically alter the auto industry, along with the entire service business and the likes of insurance companies.
Fewer cars travelling more efficiently would change the face of our cities, reducing the demand for parking spaces, for instance, and sparking a reuse of land now dedicated to cars.
In short, the world would look much different.
Whether the pros, of which there are many, outweigh the (temporary) cons remains to be seen.