A driver sits engrossed in her laptop screen, catching up on emails as the car barrels down the highway. In the next lane, a father helps his kids finish homework while their vehicle swiftly changes lanes. Nearby, an empty car returns home after dropping off its owner.
These are the self-driving cars in which humans can be mindlessly commuting in as few as five years, some ambitious estimates claim.
"It's a highly disruptive technology that's coming on a lot faster than people expect," says Barrie Kirk, executive director of the Canadian Automated Vehicles Centre of Excellence. He helps governments and companies prepare for the advent of automated vehicles.
Many automakers and tech firms have already entered the driverless car manufacturing game. Now it's a race to perfect the technology and start selling these Knight Rider-style vehicles.
Companies hype the cars as the best safety feature since seatbelts and airbags, but there's a sense that phasing driverless cars onto public roads may be anything but a smooth transition.
Humans make 'poor drivers'
Self-driving car advocates, like Kirk, believe in the technology's potential to save thousands of lives.
"Humans, generally, are poor drivers," he says. He would like to see human drivers banned from roads to make room for an all-automated-vehicle world.
'If you're got a whole bunch of sensors that give you a 360-degree scan, 30 times a second, humans can not come anywhere close to that.'- Barrie Kirk, driverless-car advocate
Drivers' mistakes are responsible for more than 90 per cent of crashes, the U.S. National Highway Traffic Safety Administration found. Kirk hopes automated vehicles can eliminate 80 per cent of such collisions — a number often cited by advocates.
In 2012, 2,077 people died in car crashes on Canadian roads, according to Transport Canada. If Kirk's estimate holds, about 1,500 of those victims could have avoided an accident.
"If you're got a whole bunch of sensors that give you a 360-degree scan, 30 times a second," he says, "humans can not come anywhere close to that."
Test cars 'far inferior' to novice driver
Today's automated vehicles don't have that capability, says Steve Shladover. He's researched driverless cars for four decades, most recently as the program manager, mobility for the University of California's PATH (Partners for Advanced Transportation Technology) program.
The steering wheel of this Rinspeed Budii self-driving electric city car can be used to drive from either side. When it's not being used, it rests in the middle of the dashboard. (Reuters)
There are five levels of automated vehicles. They range from cars with adaptive cruise control that adjust a car's speed while a driver steers, to the highest echelon, dubbed "full automation." That car doesn't need a driver. It is the driver.
"There's nothing that's even remotely approaching the ability to do that," says Shladover, whose estimate pegs such a car's unveiling to be decades away. "Even the most sophisticated of those test vehicles is far inferior to a novice driver."
Deaths and injuries are "remarkably rare" compared to how much people drive, he says. In 2012, six people died in car crashes in Canada for about every one billion kilometres vehicles travelled, according to Statistics Canada. The U.S. figure is slightly higher, according to the NHTSA.
It would be "extremely difficult" to design an automated vehicle that can run that long without getting into a serious crash, he says.
"Think about things like mobile phones and laptop computers ... They don't run nearly that long without failures," says Shladover. "But we're expecting a car to now operate that long without a failure in a very complicated environment?"
Could road safety worsen?
Even when a company successfully creates a fully automated car, there's skepticism about how well it can manage that complicated environment peppered with pedestrians, cyclists and human drivers.
This eight-passenger shuttle moves autonomously at up to 20 km/h. Steve Shladover says people can expect to see this type of vehicle operating at low speeds in small areas in about five years. (Edgar Su/Reuters)
It is "not a foregone conclusion" that a self-driving car would perform more safely than an experienced driver, claims a new report by researchers from the University of Michigan's Transportation Research Institute. It's unclear whether human or machine would respond faster and better to jaywalkers, mechanical problems, potholes or bad weather.
"There's an awful lot of communication that happens by eye contact," says Shladover. Pedestrians often glace at drivers' eyes to ensure they're seen before stepping off the curb. "You can't do that if you don't have a driver in the vehicle."
Plus, it's difficult for a computer system to recognize all potential hazards. When a ball bounces onto a residential street, it's not necessarily a problem. But when a child chases that ball, the car must stop.
"Is that vehicle supposed to stop as soon as it detects that there's a ball bouncing in the road because that might be followed by a child?" he says. "There are many complicated challenges like that."
'Even the most sophisticated of those test vehicles is far inferior to a novice driver.'- Steve Shladover, University of California
As these cars are introduced onto public roads they will have to interact with human-driven cars. During that lengthy transition "safety might actually worsen," the report says. Drivers could not rely on eye contact and feedback from cars with no one at the wheel.
Even Kirk concedes that transitional period could be rough, but he blames people in traditional cars who "take advantage of the good nature of computer drivers."
A well programmed computer driver is very defensive.
"Human drivers will cut in front of a computer-driven car because they know they can," Kirk says. "It makes human drivers even more assertive, even more aggressive drivers, which is what we don't need."
'Sensible' implementation key to benefits
There will be time to adjust before the new fleet of robot cars takes over roads.
"We're not going to be in a situation where we go from no automation to fully autonomous or self-driving vehicles," says David Adams, president of the Global Automakers of Canada.
Some people already own low-level autonomous vehicles, like ones that parallel park once the driver has properly aligned it. Some U.K. cities have started experimenting with low-speed self-driving shuttles on closed streets.
Even if safety is somewhat disputed, there are other potential benefits that can make the pursuit of these cars worth it.
Seniors, disabled people and others unable to drive will gain mobility. Families may need to own fewer cars if vehicles can travel empty to pick up and drop off family members. Cities may require fewer parking spaces if cars can return home after dropping off owners.
But to see all those benefits and ensure safety isn't compromised, these cars must be carefully brought into the public realm, says Shladover.
"It has to be done in a sensible way."
Anda sedang membaca artikel tentang
'Disruptive' driverless cars may be tough on human drivers
Dengan url
http://teknounikmenarik.blogspot.com/2015/03/disruptive-driverless-cars-may-be-tough.html
Anda boleh menyebar luaskannya atau mengcopy paste-nya
'Disruptive' driverless cars may be tough on human drivers
namun jangan lupa untuk meletakkan link
'Disruptive' driverless cars may be tough on human drivers
sebagai sumbernya
0 komentar:
Posting Komentar