Transport Committee on March 28th, 2018

[John Wall]

Yes, I think that the algorithms are going to be sophisticated enough to understand black ice, slippage of the wheels, etc. They'll be able to identify more quickly than you what's happening to the vehicle from a geometry perspective.

It's also one of the reasons that I believe autonomous driving is further out than, say, Uber would like to claim. Uber has a business reason for wanting autonomous drive. It's their business model. They don't survive without it.

There's very little testing in Canada for autonomous drive. We have an autonomous car at QNX, and in the winter it behaves much differently on snow and ice. We're nowhere near being prepared to handle those situations. I think people are still better at handling those situations, but I think the technology, with machine learning and AI, will be able to react appropriately to those types of situations.

[Robert Aubin, NDP, Trois-Rivières, QC]

Is that—

[The Chair, Liberal, Judy Sgro]

Mr. Aubin, I'm sorry. We're going on to Mr. Sikand.

[Gagan Sikand, Liberal, Mississauga—Streetsville, ON]

Thank you, Madam Chair.

A witness on the previous panel pre-empted my question by describing the period when a horse overlapped with a car on the road. It's amazing that we're here talking about driverless cars.

I was reading a report by the U.K. House of Lords in 2017 entitled “Connected and Autonomous Vehicles: The future?”. From the report, the study discussed the transitional period when traditional vehicles would share the road with CVs and AVs. The report expressed concern that the effects of CVs and AVs sharing the road with traditional vehicles are not fully understood. It recommended that the U.K. government undertake mixed-fleet modelling to inform policy development.

My first question is this: what modelling work, if any, has been undertaken here in Canada to assess the impact of a mixed fleet of autonomous and non-autonomous vehicles on Canadian roads?

[Grant Courville]

That's very admirable. We wholeheartedly agree with that recommendation, because humans by definition are unpredictable. You're absolutely right. The reality is that it will be a mixed fleet, as we discussed earlier.

With regard to what John was saying earlier, the testing we do in inclement weather with all of the sensors is not so much to handle when things are going well but to handle when things are unpredictable and things aren't going well. It's to handle things we haven't foreseen and to find out how we can react to that intelligently. Can cars communicate to each other? John mentioned artificial intelligence, so we wholeheartedly agree.

Part of what we do with the sensors is to deal with obstacles that are not predicted and whatnot. That's one of the other reasons that we have a long way to go before we get to level 5. One of the things we tend to say is that there are connected cars, there are automated cars, and then there are autonomous cars. We tend to jump from connected to autonomous, but there's automated, which is a major step. You're seeing that already with a lot of automated features in the cars. The cars are getting safer, and those are all building blocks on the way to autonomous cars.

[Gagan Sikand, Liberal, Mississauga—Streetsville, ON]

To that point, how can modelling reflect human interactions?

[Dr. Denis Gingras]

HMI problems, human-machine interaction problems, are very critical. Actually, there are a lot of unconscious details when we interact between humans—for example, the body language. If you take a pedestrian simply crossing a road, he will probably look at the car and see the driver. Without any talking, he understands that the driver has seen him or her, and he will say he's confident because the car is slowing down. We don't have this kind of interaction with a fully autonomous vehicle yet, and we have no clue on how to predict it precisely.

There's a lot of model-building with psychologists in connective science, and mathematicians trying to develop driver models and HMI models in order to have safer interactions between these highly autonomous vehicles or highly automated driving functions and humans.

To give you an example, there are some German car makers who have developed graphic interfaces with a big smile on the front of the cars, showing the pedestrian that the machine has seen you, so you can safely cross the street. We are exploring all kinds of solutions in that aspect.

This is not a trivial problem. This is also one tiny problem among an ocean of problems before we reach a full, mature solution at level 5. We're not there yet.

[Scott Santens]

I'd like to give an example too. In the case of this mix of human and self-driving cars, you can imagine that the way a traffic jam can form is just through, let's say, rubbernecking. Someone will be driving by. You'll look, and you'll slow down. That will cause the car behind you to slow down, and then the car behind, and suddenly there is a big traffic jam. It just snakes its way back.

One possibility is that this will cause all the automated cars to get into that jam as well. If you have some kind of a hybrid system, however, and you still have a human driver, but it detects that the car in front of you has started to slow down, it wouldn't slow down as much as a human would, and the car behind it would be affected differently. Then you would avoid that traffic jam that would otherwise have been caused by a human driver without any kind of assistance.

There is a kind of hybrid system there that could be implemented.

[The Chair, Liberal, Judy Sgro]

Thank you very much, Mr. Santens.

We'll move on to Mr. Iacono.

[Angelo Iacono, Liberal, Alfred-Pellan, QC]

Thank you, Madam Chair.

Professor Gingras, this morning, tests were conducted in Blainville on emergency braking systems. Could you tell us more about that?

[Dr. Denis Gingras]

Thank you for your question.

I have actually been at Transport Canada's and PMG Technologies' Motor Vehicle Testing Centre in Blainville for two days. In collaboration with the National Research Council of Canada and Transport Canada, we held a workshop to create what we call a community of practice, in other words, a network composed of all Canadian stakeholders in the road transport and smart vehicle sectors, with a view to developing a national strategy in the area.

The workshop brought together experts from NRC and Transport Canada as well as from universities such as Waterloo and Sherbrooke. There were also representatives from certain organizations in Quebec and Ontario, but also from the City of Calgary, since tests are being done there.

[Angelo Iacono, Liberal, Alfred-Pellan, QC]

What did you learn?

[Dr. Denis Gingras]

This morning, we did some tests, some with dummies, to check the emergency braking systems of various models of vehicles, specifically from Kia. As we have already seen in previous tests over the last two years, the emergency braking systems on the market do not really work beyond a speed of 40 kilometres per hour. The technology has not yet been refined, and the situation is all the more dangerous because the public has not been made fully aware.

This presents a challenge for transportation regulators, particularly in the provinces. I am referring to the organizations that offer driving courses or issue driver's licences. If someone buys a smart vehicle with automated features, such as a Tesla with an autopilot system, governments need to ensure that the new owner is well informed about the technical limitations of these functions and knows when and how to use them. This is in order to keep within their parameters and to avoid dangerous situations. It is extremely important.

There is also the whole issue of developing new regulations for these new technologies.

[Angelo Iacono, Liberal, Alfred-Pellan, QC]

In terms of computer security, is any programming system safe from piracy?

[Dr. Denis Gingras]

I do not think any of the current computer systems can claim to be completely safe from a cyberattack.

As soon as a vehicle is connected and is able to communicate and exchange information with the road infrastructure, with other vehicles or with pedestrians, I believe there is a risk that the data will be intercepted and that unauthorized people will be able to control the vehicle remotely.

We have already had an example of that potential danger in the United States, two or three years ago, with a Jeep Cherokee.

[Angelo Iacono, Liberal, Alfred-Pellan, QC]

Thank you.

[The Chair, Liberal, Judy Sgro]

Mr. Graham, you have one minute.

[David Graham, Liberal, Laurentides—Labelle, QC]

Okay.

Mr. Gingras, you bring up a subject that I wanted to raise. It was exactly that Jeep that was the target of a computer attack in 2015, a zero-day exploit, as it is called.

These are fly-by-wire vehicles, and you can take control of them remotely and take them off the road. What are we doing to prevent that from happening in the future? If you jailbreak a QNX vehicle, what are the consequences of that?

[John Wall]

Specifically on the Charlie Miller hack that we're talking about, we're pretty intimately familiar with that hack.

[David Graham, Liberal, Laurentides—Labelle, QC]

Was it a QNX vehicle?

[John Wall]

It was a QNX-based vehicle, and the way I would describe it is that somebody left the door wide open. There was no security in mind at all in that vehicle. It was a wake-up call to the auto industry that a lot more had to be done.

[David Graham, Liberal, Laurentides—Labelle, QC]

But anytime you have a network vehicle, that's going to be in the case anyway—

[John Wall]

No, no, there was very serious things that were done in this vehicle to leave the door open. I'm talking scripts that say “update me” with this. It was ridiculously open.

[David Graham, Liberal, Laurentides—Labelle, QC]

That says something to the major point here, which is that no program is better than the programmer who wrote it.

[John Wall]

Yes, there is programming and then there is the methodology of how you put the system together, and cybersecurity is never going to be an absolute certainty. It's a cat-and-mouse game in which you will always be staying ahead in a technology, because we've seen vulnerabilities in things like OpenSSL Heartbleed that happened. You don't know they're there.

We've had just recently Spectre and Meltdown with hardware. I think the—