Home, James

Self-driving cars, just around the market corner, will make drivers redundant and roads less dangerous. But how will they affect the economy?

Picture this: you're driving to work, responding to and sending emails, reviewing the presentation you’ll soon be making and looking out your window every now and again to take in the countryside on the congestion-free commute downtown. Once you arrive, you dispatch your driverless car back home to take the kids to school before sending it off to collect your parents and drive them to their doctors’ appointments. Or, if you’re entrepreneurial, you can put it to work by sending it to a pool of self-driving cars where anyone can call up and pay you a fee to use it before it returns to your office and brings you home.

This is not the stuff of science fiction anymore. It’s happening. In fact, self-driving cars will be here a lot sooner than most people think because the technologies needed to fully automate driving are coming onstream. Within three months every Tesla vehicle will be capable of driving down highways automatically, unassisted by humans. In 2013, Mercedes test-drove an autonomous S 500 through 100 kilometres of city and highway. GM’s Super Cruise technology will keep a car in its lane at a safe distance from other vehicles, adjust steering and apply the brakes. It will also "talk to" similarly equipped vehicles and share speed and road condition information.

Governments have been busy, too. The UK national budget for 2015 has dedicated £100 million ($180 million) for autonomous vehicle development. This year, four cities in the UK will host driverless vehicle trials, including testing a 10-passenger autonomous electric shuttle. Ontario’s Ministry of Transportation has partnered with the Ontario Centres of Excellence to create the $1-million Connected Vehicle/Autonomous Vehicle Research Program aimed at helping businesses and academic institutions develop and commercialize connected and autonomous vehicle technologies. Washington, DC, Florida, Nevada and California have passed legislation to permit testing of autonomous vehicles.

Not surprisingly, global financial services firm Morgan Stanley issued a report in January that predicted the end of the auto industry as we know it and outlined some big financial gains when the technology is fully adopted. Autonomous Cars: The Future is Now contends that autonomous cars could help save the global economy more than US$5.6 trillion a year and US$1.3 trillion for the US alone.

When is this likely to happen? "We will see fully autonomous cars on the road between 2018 and 2020," says Barrie Kirk, executive director of the Canadian Automated Vehicles Centre of Excellence (CAVCOE). "By 2030, the majority of cars on the road will be capable of self-driving."

So what are self-driving/autonomous/automatic cars, exactly? Any car that can drive itself without human assistance is a self-driving car but there are many shades of gray. In the US, the National Highway Traffic Safety Administration has defined five levels of vehicle automation from 0 to 4. Here’s the breakdown: Level 0, no automation; Level 1, at least one specific control function, such as pre-charged brakes that can help you stop quickly; Level 2, at least two functions designed to work together, such as adaptive cruise control in combination with lane centring; Level 3, the driver is able to drive without using hands or feet but is still expected to be available to take control — the Google car is an example of limited self-driving automation; Level 4, no human intervention necessary.

"You can already buy cars that are partially autonomous," says Kirk. "Intelligent cruise control, braking for pedestrians and self-parking are common options on many of today’s automobiles." Most automobile manufacturers have indicated that they will have self-driving vehicles in their 2020 to 2025 lineups.

We have five to 10 years to come to grips with the future, though in this time frame driverless cars for the masses will not be as fully autonomous as the trucks that Rio Tinto operates at its Pilbara mines in Australia. There, a fleet of trucks operates 24/7, year round. Monster trucks that can carry 290 tonnes of ore and overburden are controlled 24 hours a day by a supervisory control computer located at a remote operations centre more than 1,300 km away in Perth. Closer to home, Suncor is researching fully autonomous dump trucks for use in Alberta’s oilsands. But fully autonomous consumer cars will likely be very expensive at first and will probably require that numerous legal issues be addressed.

Google now has a fleet of autonomous Prius vehicles as well as a Lexus version that keep accumulating miles. At this point, their drivers still have to intervene every several hours or so. Each time the human driver feels the need to take over, Google’s team records that information and analyzes why the driver became uncomfortable. The team then programs in additional aspects of autonomy to deal with that situation the next time it happens. The tech giant plans to release a fully autonomous car between 2017 and 2020 for short hops in urban centres. "Google released a video this year where the car had automatically detected all the traffic cones around a construction site, identified new lanes on the fly and drove as if nothing out of the ordinary had happened," says professor Steven Waslander, director of the Waterloo Autonomous Vehicles Laboratory (WAVELab).

WAVELab is working on projects focused on collaborative driving where robotic cars communicate important information about their current driving conditions — intended travel speeds and routes — and join into platoons on highways, positioning themselves in close proximity to eliminate traffic congestion. Waslander and his colleagues are also working to expand the set of situations where we can drive autonomously — for example, moving beyond lane detection so vehicles can automatically handle multiple lanes, intersections and off-ramps. This past winter they looked at the effect of snow on road conditions and lane detection in order to facilitate autonomous driving in adverse weather conditions.

How self-driving cars work

An array of cameras, laser scanners, sonar and radar collect data all around the vehicle at least 30 times a second. Inertial measurement — which refers to a set of about nine sensors that includes gyroscopes, accelerometers and magnetometers that measure the rate of rotation, acceleration and the direction of the magnetic field — computes the vehicle’s orientation and the rate of change of that orientation. All those cameras and sensors are connected to a computer that fuses the data to create a 3D computer map of the area around the vehicle. Algorithms and software help the vehicle/computer figure out what to do next — stop, change lanes, adjust speed, drive — and share information with other automated vehicles to create the most accurate, real-time picture of what’s happening on the road. Underpinning it all is a GPS system that sets the start and end points of the trip.

Of course, the last big piece is execution. A team out of Stanford University has been racing Shelley, its self-driving race car (an Audi TTS), at Thunderhill Raceway north of Sacramento. A collaboration between Stanford’s Dynamic Design Lab and Volkswagen’s Electronics Research Lab, Shelley hits speeds of 120 m.p.h. while software tells it when to brake, turn and accelerate. "They are driving at the limits of friction and have reached the point where they can match an amateur race-car driver around a track," says Waslander.

Societal impact

Automated Vehicles: The Coming of the Next Disruptive Technology, a report from the Conference Board of Canada in collaboration with Calgary’s Van Horne Institute and CAVCOE, calls on governments and business to start planning because automated vehicles will change just about everything: infrastructure needs, the nature of jobs, the economy and healthcare. The report anticipates benefits to be as high as $65 billion largely because of significantly fewer traffic collisions, which immediately translates into healthcare, legal and auto-repair savings.

The figures are even more staggering in the US, where traffic accidents alone cost almost US$900 billion annually. This may be why the US seems to be preparing more aggressively for a future with driverless cars than Canada, where only Ontario appears to be working on legislation to permit testing of self-driving vehicles.

In the US, 17 states have considered self-driving car legislation. The laws that have been enacted include provisions that view the "operator" of the vehicle as the person who engages the autonomous technology when there is no one in the vehicle. These laws seem to envisage scenarios where the car could be sent to pick someone up, or return home on its own rather than sit in an unsecured parking lot for a long time, such as at the airport.

Ultimately, the technology itself will help decide how the law evolves and how effective the law is in dealing with self-driving vehicles.

a woman inside a car

Safety

Ninety-three percent of traffic collisions involve human error. Computer-driven cars can scan 360 degrees around them, at least 30 times a second, with no distractions. They are more conservative and therefore can be much safer. When car-to-car alert systems are incorporated, automobiles will be able to communicate dangerous situations to vehicles around them as well as those following. "This will take incredible pressure — both financial and in human terms — off the healthcare system, allowing for a reallocation of resources for hospitals," says Peter Wallis, president and CEO of The Van Horne Institute. "Driverless cars take the most unpredictable part of the drive away — the human element."

There is also a scenario being posited that the move to driverless cars will result in more car sharing and fewer cars on the road, leading to less congestion. "There is a general recognition that widespread adoption of driverless vehicles will lead to the emergence of transportation as a service, meaning more people will use cars for a single trip as opposed to owning them," says Kirk. "Think about how disruptive Uber has been and then take the driver out of the equation." (Uber is the six-year-old online cab company that has regular car owners and professional taxi drivers providing transport services on demand. It now operates in 56 countries and generated an estimated US$2 billion in gross annual receipts last year.) On the other hand, there may be an increase in the number of cars owned; one you are willing to share and another, your Ferrari or Aston Martin that you keep for yourself. Another alternative would be time-sharing of automobiles, or Uber for cars: simply use your smartphone to target an available "share ride vehicle" near your location and hail it.

It’s no surprise Google and Uber have both made public how attractive self-driving Uber cars would be. The same goes for the trucking industry. It will be much less costly and more efficient to transport goods with driverless trucks because up to 60% of the expense is the driver. This in turn will accelerate the growth of online shopping and change the way we shop thanks to extremely small and efficient robo cars. Waslander anticipates a much larger range of vehicles on the roads. "Suddenly you don’t need space for humans to sit. You can customize vehicles based on what they’re transporting. We’ve already seen it with drones."

As with any disruptive technology, there are positives and negatives. Commercial drivers will be the first and most obvious workers affected, but there will also be less demand for auto body shops, insurance claims adjusters, insurance underwriters, law enforcement officers, lawyers and the court system and accordingly less need for educational courses and certifications in these areas. Eventually, as the number of devastating automobile accidents decreases, the need for trauma doctors and nurses will also be reduced, as will the demand for physiotherapists and long-term care workers.

Professionals designing and implementing software and incorporating other technology solutions into self-driving cars will require extensive standards and assurance. Robotic standards will need to be developed on a global basis to ensure automobile interoperability and interconnectivity. US President Barack Obama recently encouraged the auto industry to develop such standards for car-to-car communications.

Governments and educational institutions will have to plan carefully for a potentially smaller and more competitive labour market — one that demands more specialized skills. For example, mapping will be a growth area as the need for highly secure and accurate mapping data becomes a must. With increasing connectivity, the ability to keep cars safe from hacking attacks will be critical. "Highly specialized service, maintenance and systems-integration businesses for these vehicles will emerge," says Matt Rendall, CEO and cofounder of Kitchener, Ont.-based Clearpath Robotics.

"There is a popular school of thought that the automobile of the future will be the largest consumer electronic device you own. In an extreme sense, a self-driving vehicle is a smartphone on wheels," says Rendall.

In this near-utopian society with driverless cars making us more productive, less stressed and less likely to be involved in an automobile accident, there will still be people who enjoy being at the wheel on a sunny day, with the top down and a winding road ahead — just for the thrill of driving. Thankfully, the chairman of Audi envisions a future where we can still drive for fun but go into automatic mode for commuting. That sounds pretty good.

CANADIAN INNOVATION

Ottawa-based QNX Software Systems, a subsidiary of BlackBerry, produces a range of connected-vehicle software solutions that autonomous vehicles need to operate safely. "Connectivity to the outside world through mobile devices and built-in modems, as well as between sensors and microprocessors within the car, will play a key role in autonomous vehicles," says Andrew Poliak, the company’s global director of business development. At the 2015 International Consumer Electronics Show in Las Vegas, QNX presented two concept vehicles — a Maserati Quattroporte GTS and a Jeep Wrangler — that integrated third-party and QNX technologies, including sensors, cameras, navigation engines, cloud-based services, speech interfaces and acoustics software. In the Maserati, QNX technology connected the instrument cluster to the radio and entertainment system so that when you access a map, the cluster can give you a turn-by-turn indicator of your next manoeuvre. In the Jeep, QNX showcased how its technology works with advanced driver assistance systems to get information from street signs. "We married a navigation solution with predictive information on the route to a vision algorithm that could detect a street sign, identify the speed, compare the two and alert you if they were different. These are capabilities we already support," says Poliak.