The drive to bring autonomous vehicles to market has accelerated over the past decade, but that doesn’t mean that self-driving cars will really be roadworthy anytime soon.
“We are far from the level at which a machine can drive like a human in all conditions,” said Amir Khajepour, professor of mechanical engineering and mechatronics at the University of Waterloo.
“I’m not sure if that will happen in my lifetime.”
In 2016, Ontario started its automated vehicle pilot program. It recently extended guidelines being able to test some self-driving vehicles on certain roads without the help of a person behind the wheel.
The changes allow a passenger in the vehicle or a remote control, but there are still major hurdles to overcome.
The biggest, Khajepour says, is making machines that mirror how the brain works while driving, and that’s far from easy. Add in discussions about insurance and ethics, and it’s clear the industry is so much more than just technology.
“When we drive a car, we draw from many other experiences that we have in our lives,” said Khajepour.
Halloween, Christmas, snow under the challenges
Programming robots to respond to real-world situations has been challenging and inherently safe, says Steve Waslander, associate professor at the University of Toronto’s Department of Aerospace Studies who specializes in AI perception .
“Where they still seem blocked are … all the rare occurrences, the strange situations, the infinite variety of the human world,” he said.
CLOCK | Building machines that mirror how the brain works while driving is still the biggest challenge, says Professor:
One example of this, Waslander says, is that self-driving cars can have difficulty recognizing children in Halloween costumes. To improve on this, Google’s self-driving car project, a subsidiary called Waymo, recently had its employees’ children paraded in front of their cars in their Halloween costumes so its sensors could gather more information.
Christmas lights can also be a challenge Winter weathersays Waslander.
In addition, autonomous cars are not yet equipped to drive like humans, who sometimes give other drivers subtle cues to signal their intention, such as turning right at an intersection.
“It’s amazing what we’re doing in parallel as we move through the area and it was a lot more challenging than we expected,” said Waslander.
Electrics, Ethics, and Insurance
But technology and software aren’t the only considerations for the industry; There is a whole “ecosystem” of factors to be considered, as Khajepour puts it. Engineers have the additional job of debating the ethics of creating artificial intelligence and finding out who is responsible if an autonomous vehicle crashes. Although it has been rare so far, it does happen.
On December 16, an autonomous bus in the Durham area. That makes a four-mile loop that ends at Whitby GO station, went off the road, and hit a tree. A 23-year-old male nurse suffered life-threatening injuries and was admitted to a trauma center for treatment.
The accompanying person was the only person in the vehicle at this point.
Nevertheless, in such situations the question arises of who is responsible in the event of an accident.
“They are looking for ways [like] Extending flawless insurance to autonomous vehicles or by treating the autonomous vehicle like a driver, “Waslander said.
Because of these debates, we are likely to see self-driving vehicles in controlled environments, such as on campus, in shopping malls and warehouses, including the University of Waterloo, in the near future.
Khajepour and his team developed the technology of the WATonoBus, a self-driving shuttle that runs a 2.7-kilometer loop around the campus at a top speed of 20 kilometers per hour.
A driver is still on board during the test drives, but the team is waiting for provincial approval to start driverless tests as part of the Automated Vehicle Pilot Program.
Khajepour and Waslander both say that we might see self-driving trucks on freeways sooner than cars on city streets because there are fewer unpredictable factors such as pedestrians and cyclists to consider.
Raed Kadri, vice president of strategic initiatives and head of the Ontario Vehicle Innovation Network (OVIN), which connects researchers with industry, suggests that there’s one more factor: electric vehicles that pose different challenges.
A 2020 paper published by Carnegie Mellon University researchers found that driverless cars use more energy than cars with drivers, reducing range and requiring more charging. Researchers are now investigating how drivers perceive the reduced range.
“As we move into electrical engineering, this will be the new platform on which we will add new technology – new automated functions and connectivity functions,” said Kadri.