THE STANDING SENATE COMMITTEE ON TRANSPORT AND COMMUNICATIONS
OTTAWA, Tuesday, May 16, 2017
The Standing Senate Committee on Transport and Communications met this day at 9:30 a.m. to continue its study on the regulatory and technical issues related to the deployment of connected and automated vehicles.
Senator Dennis Dawson (Chair) in the chair.
The Chair: Honourable senators, this morning the committee is continuing its study on the regulatory and technical issues related to the deployment of connected and automated vehicles.
Appearing before us, we have Denis Gingras, Professor, Laboratory on Intelligent Vehicles, University of Sherbrooke, and Tony Zhijun Qiu, Professor, Faculty of Engineering, University of Alberta. Last year Mr. Qiu hosted the members of the committee during a visit to the University of Alberta. I would like to express my sincere thanks for having done that.
Welcome everyone. I invite Mr. Gingras to make his presentation, and I will then give Mr. Qiu the floor.
Denis Gingras, Professor, Laboratory on Intelligent Vehicles, University of Sherbrooke: Honourable senators, I would first like to thank you for inviting me and giving me the opportunity to express my views on this subject.
As you know, mobility is extremely important and key to the development of our society and civilization. That is something we will all agree on. However, paradoxically, the road transportation system, as we know it today, is quite inefficient, and that is somewhat of a surprise. If we examine the different parameters, we realize that it is an extremely inefficient system. First, let us consider the business model and the economic system, which are based on individual ownership of vehicles. There are definitely too many vehicles on the road. I believe that road transportation agencies all agree on that. Then there is demographic growth and urban concentration. This translates into congestion and overloading of our road infrastructure.
Additionally, the transportation of merchandise has migrated from rail to road infrastructure in order to satisfy “just-in-time” market demands, thus increasing the stress on our road infrastructure.
Vehicles are used on a single occupancy basis 80 per cent of the time, which means that the effective payload of vehicles is about five per cent based on the total weight of a vehicle. That is not really very efficient. On average, vehicles are used one hour per day; for the remaining 23 hours they are parked, rusting, and useless. Vehicles are still equipped with combustion engines that use gasoline and fossil fuels and have an energy efficiency of roughly 25 per cent to 35 per cent. That is not very good. Furthermore, vehicles are still being driven by humans, who are responsible for over 90 per cent of all accidents on our roads.
In the 100 years since the invention of the automobile, incremental improvements have been made. However, most vehicles on our roads today still have no awareness, intelligence, or sight. Fortunately, there has been a technological revolution over the past two decades. I have been working in this area for 35 years.
I would like to talk to you about three things: The first is the electrification of a vehicle’s powertrain, which mainly has an environmental impact, but does not change the paradigm of our land transportation; the second is driving automation; the third is connectivity.
These three main areas of technological change can result in new solutions which, we hope, will improve the efficiency of our land mobility. We would like vehicle automation to ultimately result in completely autonomous vehicles so we can eventually migrate from a system of individual ownership to a mobility services model. This would allow us to substantially reduce the number of vehicles on our roads.
To enhance road safety, we would then be able to take the driver out of the equation and reduce the number of collisions and fatalities.
There will be major impacts, not just in terms of technology, but also on insurance companies, legislation, the after-sales and maintenance markets, and everyday life, such as the need to obtain a driver’s licence, vehicle registration, and so forth. It is really going to change everyday life.
This will also advance the “last kilometre solution”, which results in the better integration of different modes of transportation and improves the efficiency of the interface between different transportation systems which, at present, work in silos and independent of one another.
However, we are a long ways away from having fully automated vehicles because of the enormous challenges yet to be overcome. The safety of an automated vehicle is based primarily on two main parameters: robustness and reliability. These are two entirely different things. With respect to robustness, an automated vehicle is a robot on the road that operates in an open environment, and not just because of mobility. Typically, the environment of a vehicle travelling at 100 kilometres per hour changes every three seconds. That is already enormous and requires real-time applications and systems that perform extremely well and have very rapid response times.
There are also many other parameters including the weather, and traffic and road conditions. The result is a multitude of driving scenarios that cannot all be foreseen or anticipated. An automated vehicle cannot be programmed like a traditional system, as we know it today. To handle all these variables we will need an extremely robust system.
Then the complexity of the systems comes into play. Today, a vehicle typically has more than 40,000 parts and components, requires more than 100 million lines of codes, and uses about one hundred microprocessors. This requires validation and testing methods that are again altogether different than those we currently use. It requires techniques based on what are known as stochastic algorithms, or sampling methods, somewhat like those used by major industries in microelectronics, for example. We cannot conduct a deterministic and systematic verification of all the system’s possibilities. We have to use sampling and, at best, obtain an idea of the system’s performance on the basis of probability.
This leads us to a major challenge, which is artificial intelligence. You may have noted that in the past eighteen months major automobile manufacturers have invested billions of dollars specifically in artificial intelligence. That is not surprising given that the most significant technological challenges concern artificial intelligence. We are a long ways from having robots with cognitive abilities, perception, awareness, and decision-making capabilities equal to those of a human being. The strengths of computers are speed, logic, and combinatorial computing. However, human beings can do things that machines are not yet capable of. The proof is that when a human being is born, it takes years to develop motor skills, cognitive abilities, perception, and awareness of self and their environment. That is why we issue a driver’s licence only to a young adult who is at least 16 years old, and not to a five-year-old child.
Here are a few observations about connectivity. Generally speaking, connectivity allows for the exchange of information, which can be used for two purposes: first, to improve the redundancy of information available to smart vehicles; second, to provide information that is complementary to the information available in intelligent vehicles. The redundancy is important because it allows us to validate the information that we already have, increase our confidence in this data, and thus improve the probability of good decision-making in any driving situation, for example, when avoiding a collision or an obstacle.
Having complementary information allows us to expand perception, namely by knowing that certain information can be obtained by means of communications beyond the distance at which our own on-board sensors can perceive things. Typically, the on-board sensors of a vehicle have a maximum perception of approximately 100 or 200 metres at best. By using telecommunications, we can seek information from sources that are much further away.
Communications allow us to envisage and design mobility solutions where the intelligence would be distributed and no longer concentrated only in the vehicles. At that point, we could exchange information between road infrastructure and vehicles. For example, if a city has some intersections where the accident risk is greater than in others, these intersections could be equipped with instruments, made intelligent and thus enable vehicles, sometimes less intelligent ones, to obtain information and make appropriate decisions to avoid accidents.
Here are some recommendations. First, I believe that we must develop strategies, in Canada and elsewhere, because this is an international and not uniquely Canadian problem. Governments, automobile manufacturers, road transportation agencies, and research establishments must put in place a global development and deployment strategy in keeping with the concept of smart cities.
With respect to the economy, the fusion of the automobile sector and information technologies opens the door to many business opportunities for Canadian companies, and the development of a new generation of companies, start-ups, SMEs, which will be multidisciplinary in nature and will combine information technologies with the mobility context.
Finally, as a professor, I believe that there must be a complete overhaul of the curriculum of our education system at the college, professional, and university levels so that we can train a new generation of professionals and highly skilled workers who will be able to meet the challenges of complex systems and multidisciplinary problems.
The Chair: Thank you, Mr. Gingras.
Professor Qiu, your turn to enlighten us.
Tony Zhijun Qiu, Professor, Faculty of Engineering, University of Alberta, as an individual: Thank you for the invitation. I’m very glad to be here to present what we have done and show some opinions we have on connected and automated vehicles.
First, from a technical point of view, I want to clarify the concept for connected and automated vehicles. Automated vehicles is very clear; it is vehicle technology. I don't know if we are familiar with a connected vehicle; sometimes we are confused.
Basically, a connected vehicle and, in my opinion, is more likely trying to digitize our infrastructure and make it compatible with the vehicle technology change.
A connected vehicle we called previously the vehicle infrastructure integration. I think that’s a pretty good term, for different reasons for converting a connected vehicle.
We can see how we can use ICT, information and communications technology, to digitize our infrastructure. With an automated vehicle, the first step forward is definitely how to apply the ICT technology to make our vehicle smart.
The second point I want to say on connected automated vehicles is that last September when this committee visited my lab, we spent around 40 minutes at a demo in Edmonton. As I planned, a connected automated vehicle, it tried to provide integrated, open and data-driven platform for a future transportation system. For example, in different municipalities, we have a challenge. The city council received a proposal from transit. City council received the proposal for fire vehicles. City council received a proposal for winter maintenance cars. Actually, with all that talk about infrastructure, they separated the proposals.
When I talk with some government officials, they say, “Oh, why are these things so disputed?” They don’t know how to integrate.
A connected vehicle used that as a way and then tried to make our infrastructure digitized and essentialize the management so that our different vehicle units can be connected with the same platform. That’s why we called it an integrated, open and data-driven platform. Also connected vehicles is a way we have to make efficient use of the existing invested infrastructure. When we get into questions, we can talk more about that.
Also, these connected automated vehicles is a way we can leverage with technology booming and from IT and the automobile industry.
Definitely one major principle is to try to reduce the error from human beings.
In Edmonton, since 2012, we have worked together with University of Alberta and the University of British Columbia and with the City of Edmonton. We installed 30 roadside equipments and installed quite a few vehicles and tried to research and develop and demo this technology and how it can help our federal and provincial governments to operate traffic in the future.
If you didn’t attend last year, any time you’re welcome to visit Edmonton, and we will show you what we are doing now.
Finally, I will mention how a connected automated vehicle is unique for Canada. First, in Canada, cold climate definitely is a major issue. I don’t know how other countries are doing it. In Edmonton, in the cold climate, we are testing how well it is working at minus 30 degrees. We didn’t test at minus 40, but minus 30 last year, and we collected a lot of very good evidence for future policy and related studies. We have a long winter, so that is unique in Canada.
Also, urban city-oriented traffic, because our traffic is more like city traffic, and also compared with other developing countries, we have some unique situations as well.
The last one is infrastructure investment on public transit. In Canada we spend a lot of money on the roads, and we have a lot of work to do to think how we can make efficient use of the existing infrastructure. For example, we talk about connected automated vehicles to help increase the capacity. For example, we talk about 1800 vehicle power, and a connected automated vehicle possibly can be increased to 2100 power or 2400 power. In that sense, we need to spend money in a smarter way.
Looking at future steps, first, and definitely very important for regulation, a connected vehicle, and also, we need to have some policy on radio licence, and a Canadian Standards Association, CSA, approval, because our infrastructure has been digitized. We need to get approval from CSA.
We need to have a very reasonable and proper advocate for automated vehicles, because automated vehicles were driven by industry for a long time, for the past few years, but what is the government's role for automated vehicles from a policy point of view, and also we need to highlight how we can make our infrastructure ready or better to make it compatible with this vehicle technology change.
Last one, also the most important one, how is Canada positioned in the future with the connected automated industry market?
We have some concerns about job creation. We have some concerns about jobs in relation to automated vehicles. We shouldn’t worry too much. Previously we talked about how a computer will kill our jobs. But at the same time, because of computers, it generated new jobs as well.
As Denis mentioned, we need to think in advance how we can compute our education system to train our next generation for our infrastructure. Thank you.
Senator Saint-Germain: Thank you, and congratulations to both of you. As witnesses, you have provided information that is of great interest.
Professor Gingras, you spoke about the importance of global or comprehensive strategies. In particular, you mentioned the importance of changing the curriculum. My first question about comprehensive strategies is related to your reference to the “last kilometre solution”. Could you tell us a little more about your vision for the harmonization of individual transportation and public transportation and what planning needs to be done to accomplish this?
Mr. Gingras: That is a very good question. It is interesting because just in recent weeks the media have been reporting on agreements between certain municipalities and the company Uber, not to name names, to provide public transportation options for small municipalities. Traditional public bus services are sometimes too costly for small municipalities.
The “last kilometre” is one of the key problems. I worked for almost 10 years at the Institut national d’optique in Quebec City, in the field of optics and photonics, on transportation applications and the development of fiber optics. We also looked at how to solve the problem of the “last kilometre” in order to foster the transfer of general information to residences.
There is a fairly similar problem in the context of physical mobility. I believe that Uber vehicles are currently a type of mobility service that is less expensive than conventional mobility services such as taxis. Ride-sharing is becoming increasingly popular, especially among younger generations.
I am thinking of autonomous vehicles that serve as shuttles, for example, and can transport a certain number of people. We see this in developing countries, where community taxis can accommodate five to ten people in one vehicle. These shuttles make local trips of a few kilometres. I believe that this type of solution is very flexible and would resolve to a great extent the problem of the “last kilometre” by moving people to access points for conventional public transportation networks, such as commuter rail lines, subways, express bus networks, and so forth.
I believe that autonomous vehicle technology is going to play an important role in this regard and will result in a significant reduction in the number of vehicles on our roads because many people will no longer find it necessary to buy a vehicle, especially in major urban centres.
Senator Saint-Germain: Professor Gingras, you aptly mentioned the example of Uber. You are a member of the academic community. The concern we have is planning for and supporting this. We are under the impression that there are various research studies, various teams, and various isolated but successful efforts. However, in terms of the appropriate use of a comprehensive public transportation strategy for Quebec, and also the country, is there a pilot project underway somewhere? If not, what can be done?
Mr. Gingras: In fact, reality is catching up with this aspect. I am part of a group, together with the Quebec transport ministry, which is studying the issue. I am not an expert in all areas, but I have learned a great deal about the complexity of organizational infrastructure and municipal, public, and private transportation agencies. There are also those who look after sections of highways, federal bridges, and the provincial transportation department. These organizations generally are still not exchanging information. At this point, data banks do not talk to one another, and optimizing traffic for the current road infrastructure is not done globally because the road infrastructure is segmented and the responsibility of several entities. For now, that is where we find ourselves.
At this time, there is the political will to try to solve this problem. However, we are a ways from a global solution and a well-structured plan for the deployment of new technologies such as autonomous vehicles. I believe that the manufacture of autonomous vehicles is going to require federal legislation and collaboration with the provinces, the United States, municipalities, and the agencies responsible for road infrastructure, because I am convinced that in order to come up with solutions for intelligent mobility, we are going to have to have intelligent infrastructure as well as intelligent vehicles. It is true that vehicles, ideally, should be completely autonomous. However, before we get to that stage — 20, 25, or 30 years from now — we will have to find solutions much more quickly in order to solve our dramatic road transportation problems.
Senator Boisvenu: First of all, I would like to apologize for being late.
My question is for you, Mr. Gingras. I am from the Eastern Townships and am pleased to be speaking with someone from the University of Sherbrooke. I would like to remind my colleagues that if we are able to use this today, it is because of the research by the University of Sherbrooke, which invented this famous chip that is now found in many cell phones. I would like to congratulate the University of Sherbrooke, because I am very familiar with its accomplishments in the area of research and training for our Quebec scientists.
My first question concerns the fusion of the traditional with the new, and what lies ahead. In your research, are you strictly focused on new technologies, without considering how this new technology will co-exist with traditional transportation? I believe that when we moved from horses to motor vehicles, this transition did not happen overnight, but took several decades. Have you thought about how this technology will align with a relatively long transition phase?
Mr. Gingras: Yes, we think about that a lot. We are trying to find solutions. The problem cuts across many areas. Eventually, there will be hybrid traffic where the majority of vehicles are still driven by human beings, but where we will increasingly integrate vehicles in automated driving mode and vehicles that are completely autonomous. We do not yet have clear answers.
One of the approaches in artificial intelligence that I personally am studying in my laboratory is the Turing test. One of the fundamental questions that has no clear answer is how to define the safety metrics for autonomous vehicles. Will there be social licence for the massive deployment of autonomous vehicles on the roads because the number of accidents per millions of kilometres is lower than the number of accidents caused by human drivers? Must we consider the number of individuals who have died per million kilometres? Will probabilistic metrics or mathematical metrics be used to define the level of reliability of the global system? You can see that there are many ways to define a metric for road safety.
An approach that is more closely related to artificial intelligence is the Turing test. I don’t know whether you’ve ever heard of it. There is a movie about Alan Turing, the man who invented computers during the Second World War. He is the one who cracked the code of the German’s enciphering machine, and so on. In the 1950s, he invented the Turing test at the beginning of the computer era. In the 1960s, a program called ELIZA was developed. A person would ask questions and talk to a black box without knowing whether there was another person or machine inside the box. The box printed out answers and the person asking the questions had to determine whether the answers came from a human or from a machine. If the person could not tell the difference, the machine was considered to be intelligent. That is the basic principle of the Turing test.
Now, to adapt that to completely autonomous vehicles, we need to compare the dynamic behaviour of vehicles driven by human beings to that of vehicles driven by a completely autonomous engineered system, or an intelligent machine. If no major differences are observed between the dynamic behaviour of the two types of vehicles, then automated or autonomous vehicles would be considered to be equivalent to those driven by human beings. This approach is rarely discussed, even though this is still rather a hot research topic. Researchers are trying to determine how autonomous vehicles can be seamlessly integrated into a flow of traffic consisting mainly of vehicles driven by human beings. They are considering factors such as road rage, et cetera. Obviously, a computer does not exhibit this type of behaviour, but it comes with other issues.
Senator Boisvenu: It may be the user that is causing the problem.
Mr. Gingras: The passenger could become enraged.
Senator Eggleton: Gentlemen, thank you for your submissions.
Another area of concern, the safety of the vehicles particularly in the mixed transformation situation,is going to be of vital concern to us here and to governments in general, but safety in terms of driving is one thing. Safety in terms of the data being collected is another thing.
The United States Senate and House of Representatives both have bills before them. The one in the Senate is called the SPY Car Act of 2017, and the one in the House of Representatives is called the SPY Car Study Act of 2017, one word difference. They both get into the area of cybersecurity standards that need to be met by the industry with all the collection of this data and its value to the manufactures and the concerns about it invading people’s privacy. They have some recommendations about cybersecurity standards and some privacy standards. Are you familiar with that, and what would you say about those endeavours?
Mr. Qiu: From the vehicle manufacturer, in the past few years, they spent a lot of money on research. We have the first vehicle manufacturer to have the module-based vehicle in production with GM, and the 2017 CTS Cadillac, they have that model already. We work with GM now, and GM sends their car to Edmonton, and we prepared one car and connected it with a GM car.
The question to the City of Edmonton is how we can make our infrastructure secured. When we move to the deployment stage, subsequently it’s very important, just like all computer viruses, and subsequently technically it is not very difficult. It is like our cellphone or computer and how we protect it. And also we have our credit card and viruses that are created in finances and banking.
On the technical side, it will be easy. As to how to deploy it and what the policy is, that will be the challenge. Thank you.
Mr. Gingras: Recently I completed a study for Transport Canada on the impact of V2V, V2I connectivity in road transportation, and DOT and NHTSA in Washington are developing bills to regulate the use of DSRC, the dedicated short-range communication systems, and to oblige the automakers, the OEMs, to deploy this technology starting in 2019 for all their new models.
It’s a question of standards, basically, and I would say that Canada has not much choice in eventually following the standards that will be agreed upon between Canada, the United States and North America based on the SRC.
The SRC includes a package of substandards, different rules, concerning the construction of what is called the basic safety messages that need to be transmitted from one vehicle to another, the regulations on the different channels being used in priority of the messages and so on, and the structures that also include solutions for the authentication of the messages and to take into account the identity of the owners of the vehicles and their destinations.
The lifespan of a basic safety message is usually in the order of 40 seconds at most because of the mobility, and those messages are being used and shared between vehicles only in the immediate surroundings, in a given traffic scenario.
These elements of solutions in the standard allow us to cope with those issues of cybersecurity. There are two different keys. There is a public key and a private key system that is being implemented in the SRC.
It’s not perfect. It’s not being deployed right now. In version 1.0, there will probably be bugs and flaws, but we hope that the system will be a good start in avoiding damaging cyber-attacks in the system.
Senator Eggleton: One other one deals with the question of job opportunities or losses. You said, Professor Qiu, when the Internet came along, it did create a lot more jobs. Automation results in a lot of low-skill jobs being lost. Not every taxi driver or truck driver will become a high-tech computer specialist, so there is an issue here.
I do applaud what you both are saying in terms of the need for producing a new generation of skilled workers in this field, which is one of the areas you have suggested the government can help out in. If you have any further comments, please.
Mr. Qiu: First, the low- or high-skilled worker over time is a relative concept, just like 20 years ago, if you were familiar with a computer, it would have counted as a high-skilled worker. I think that is a technology change and how we define low and high skills. We have different definitions.
I will give one example. We are working with one high-end resort hotel group and trying to pilot-automate a chateau. The reason was to have the driver working -- but the driver is not always working -- for one day for seven hours but really working two hours. They try to reduce the human resource cost, and they try to hire some skilled technicians to do the vehicle maintenance and to provide service to the automated chateau. I think that’s why I think it will eliminate some jobs and also create some new jobs.
At the same time, to operate our infrastructure now, it is very easy to add more and more systems, but we don’t have enough employees to operate the systems. We always create new components, IT components and so forth, into the infrastructure, but our employee level is the same. We lack the human resources to operate our infrastructure.
Mr. Gingras: I could add a few comments to that. Talking about jobs in this multidisciplinary environment, the fusion between the information technology sector and automotive sector gives you one example. With respect to simulation, computer devices and simulation tools, if you compare it to aerospace and aeronautics sectors, the virtualization of the systems, modellization of the cars, the mobility solutions and the simulations of those systems will become a primary need, and those tools are essential.
For example, in the Montreal area, they are very strong in computer software and in simulation tools. You may think of CAE and the aerospace industry. I am working with a company called OPAL-RT which specializes in parallel computers and real-time simulators for simulating complex systems.
If you take, for example, the problems of testing and certification of intelligent vehicles, we have the Blainville test site facilities with PMG Technologies and Transport Canada. I know Franck was here before. I have two or three PhDs working with them, actually.
Now, just performing physical tests on test track is not enough. There are basically three approaches for validating autonomous vehicles or intelligent vehicles in general. One is the classical test on tracks. We also have what we call field operational tests. That means that like the setup that they did in Pittsburgh with the fleet of Volvo cars and Uber. You have engineers sitting behind the wheels, and you drive hundreds of thousands of kilometres. You gather data and you produce a huge database, and then you have to process this huge amount of data to excerpt the most relevant information in terms of where the systems fail and where the critical driving scenarios occurred and so on, and to build up statistics. The statistics are essential in order to build the mathematical models behind it.
To make the answer brief, a lot of the work that needs to be done will be at the mathematical, software, algorithm, simulation and virtualization levels. We can migrate a lot of work being done in the automotive sector of the future and exploit the know-how we already have in information technology to migrate into it.
Senator Bovey: I’m really interested in what you both have put forward, and I thank you very much.
I’m excited by the work being done in the private sector, at the university level, nationally and provincially, but you mentioned there is a lack of sharing of the information between one arm and another, though there is a political will. At the federal level, our job is to try to pull it together. I’d be interested in what you think we should be recommending to pool that information that’s being developed at all these various levels.
Secondly, I was very interested in your comment about education. Of course, education is a provincial responsibility. If we’re going to make sure that we are benefiting as a nation from all the research being done in various educational institutions and the training that’s going on, how do we bring all the players under one roof? How do we do that as a federal body making recommendations when, obviously, some things are highly guarded by the provinces? We’ve talked about municipalities as well. What do you think we should do to be able to make sure those playing fields are right and fair across the country and that there is knowledge-sharing and not so much double expenditure?
Mr. Qiu: I think everyone from different levels of government has the need to do knowledge transfer and sharing, and the know-how. You will find in Canada now different groups working on automated vehicles, policy and technology review.
I think in the federal budget the Smart Cities Challenge is a good opportunity and a good way to improve our infrastructure operation to a lesser level. I would suggest Smart Cities Challenge is a good way to encourage the early adoption or early pilot test of automated vehicles or connected vehicles.
For the project we are doing now, we involve three levels of government with the City of Edmonton. Now the City of Calgary has also joined. We have found in the past five years that we built very good trust. Authorization and pulling data together, and at the same time, this year GM and another four companies also joined us. This is a very good partnership. We have very good exchange mechanisms and we have regular meetings.
I think at the beginning basically with connected vehicles, we need connected people first. When we have connected people first that talk to each other, everyone will find we need to have regular meetings.
Mr. Gingras: That’s a very interesting and huge question. Probably the answer would be more at the political level than the scientist level. However, I have some ideas on that.
First, we have to find ways to increase the communication and interactions between the different levels of government. I think it’s also a question of mindset. Those problems are extremely complex, and everybody around the table should understand there is no way we are going to be able to achieve any efficient solution if everybody is trying to pull the carpet on his side and there is no consensus on the fact that each has to do its part and that we are part of a more global network, so to speak, and that we have to work together.
To change our mindsets, we have to convince people, because those organizations are directed and led by people, so if we convince those people and are able to share a vision, then maybe we can succeed.
Now, to do that, as my colleague mentioned, it may help to set up some pilot projects involving the different levels of organizations or agencies. We can envision some consortium or networks involving R&D institutions, industrial partners and public agencies from the provinces as well as at the federal level so that they can work together on common goals. In that way, they get used to working together and they can develop a shared common vision about those solutions.
These are some examples where I would start, but definitely we need to increase the efforts in talking together and trying to develop a common vision by being conscious of the fact that this solution will require efforts from all parties around the table, really.
Senator Bovey: Are universities sharing their curriculum development plans across the country? I worked in a university and I know what it is to develop curricula, and I guess sometimes we share it and sometimes we don’t. We have two prestigious universities here. Are you sharing some of the curriculum developments so that we’re building in that training?
Mr. Gingras: There have been some examples in the past that we have done. For example, in the aerospace industry, we have developed some joint curriculum at the master’s level with two or three other universities in the Montreal area. We have University of Sherbrooke, ETS, Polytechnique, and a few colleges. That works fine. We also had some joint programs in the environment.
In the automotive industry, I was one of the founders of AUTO21 NCE with Peter Frise at the University of Windsor. I was responsible for the research program on intelligence systems and sensors for 15 years. We started in 2000. And it’s funny, because at that time, when we talking about autonomous vehicles, we were preaching in the desert. Nobody heard about us. Everybody thought we were fools, utopian professors dreaming awake, and things have changed thanks to the Google demonstration in 2008-09 and now the phone doesn’t stop ringing, really.
One little message I’d like to pass on here at the same time is that we have the NCE program at the federal level. One of the basic rules is that, after 14 years, the network is dead. We kill those programs after 14 years. I was involved in the Iris pre-comp programs in the 1980s and 1990s, and then I worked 15 years in the AUTO21, and let me tell you, that hurts a lot to see very prosperous, efficient research networks across the country being killed just because of a simple rule with no justification or reason at all. AUTO21 was very efficient. More than 50 per cent of the funding came from the industry, from the automakers, from the tier companies; nevertheless, because of that rule, we killed the network. The network died last year, in April 2016, and today we need it more than ever, and now we’re discussing what we should do in Canada. We had one part of the solution. We killed it. I think this is some kind of rule that we need to change in the NCE program. I will close the parentheses. I passed on my message.
Senator Galvez: Thank you very much for these very interesting presentations. I take from both your presentations that the present transport system is completely inefficient, mainly because we waste fuel, because the cars are only used during rush hour and the rest of the time they are parked, so would this technology advance? There is a transport revolution. Want it or not, it’s going to happen anyway, and it’s going to go in the sense of electrification, automation and connectivity. My question concerns connectivity.
My understanding in reading this is that you have short-range connectivity that has launched all these advances, but very soon we are going to be connected via satellite so the perception of these automated vehicles is going to be increased. Now, saying that we want all the sectors to be implicated — the government, the municipalities, the private sector — in view of the recent hacking and ransom attacks on hospitals, I’m really worried about the fact that there is vulnerability growing because we are talking about programs that control vehicles that don’t have pilots and conduct a lot of people. We are not talking about software like Word that is going to be an open source and everybody can change it. I’m really worried about the security and safety.
You use the word “utopian” and we imagine a bus without a driver and all these people, and somebody hacking in. How can we see the future to plan, to stop or diminish this risk?
Mr. Qiu: I think about how computers and the Internet have changed our lives. Over the past few years, we started to use computers and the Internet, and we put our pictures on the Internet and also we put our credit card on the Internet to buy a lot of things, so the new technology changed our lives, brought a lot of convenience to us, but at the same time brought a lot of risk as well.
Similarly, if in the future we see automated vehicles on the road, you also think about that more like a lot of ATMs. We just put our credit card or debit card there but all our passwords are there. We drive our car, and much like we put our credit card or debit card into the ATM, how can we make it secured? Our current infrastructure is not actually as secure because we don’t have private access and we cannot use our computer to access. There are not enough people to try to use our computer to access our control and what you might design. If we know how to try, we can test our infrastructure now.
I think that just provides high requirements. We will apply our existing technology to secure our infrastructure, and also it will change our perception as well as in the future and how we need to operate our infrastructure in order just to operate on the road and also to operate our information network. So I think data in the future will also be part of our set. That’s my opinion.
Mr. Gingras: Thank you for that relevant question. Cybersecurity is always a very important issue, and people are very concerned about it. In general, it is important to remember that, any time a new technology is introduced, it can be used as a tool to improve society or it can be used as a weapon. It depends on the individual and his or her motives, whether they are good or evil. No matter what the technology, we can never completely escape that fact.
That being said, as I was explaining earlier, DSRC systems are based on standards that were developed nearly 20 years ago, which include data protection mechanisms, such as double key systems — private and public keys — or cryptography systems that protect the data and ensure that the messages sent and received by the vehicles are valid. It’s true that as soon as these vehicles become connected and can communicate with each other or with other infrastructure, they become more vulnerable in that someone could take control of the vehicle from a distance. However, with regard to the systems in the vehicle that control the vehicle’s dynamics — breaking, accelerating, speed, steering, orientation, and so on — all of that is done by isolated systems, telecommunications systems. In theory, the control systems will not be directly connected to telecommunications systems. There are firewalls.
Two years ago, an experiment was posted on the Internet wherein two engineers from Carnegie Mellon University took control of an automated Jeep Cherokee from a distance. Right after that, the American manufacturer changed and improved the design of its system to prevent this sort of thing from happening. The experiment did, however, show that a person would have to have a great deal of expertise to take control of a vehicle from a distance. It is not just anyone who can do it. It is really very complicated to do that sort of thing.
However, that vulnerability does exist. To give you an idea of how concerned auto manufacturers are about control from a distance in terms of manufacturer accountability, standards were developed over 20 years ago, in 1995-96. I remember that, at the time, everyone was saying, “In two years, vehicles will all be able to talk to and communicate with each other. The technology exists. With 5.9 gigahertz, there’s no problem. We’re going to do it.” However, it is now 2017, and we still don’t have any communicating vehicles. The reason for that is very simple. There was a major problem with the reliability and quality of the information exchanged and the vulnerability of the systems to cyberattacks. Manufacturers therefore preferred to wait for the technology to improve before deploying it.
Right now, manufacturers are also under a lot of pressure from the American government, specifically the National Highway Traffic Safety Administration, or NHTSA, and the Department of Transportation or DOT, to deploy this technology by 2019-20 and incorporate it into the new models that will be sold on the market.
I would like to add one last comment in this regard. We do not really know what is going to happen because the standard is, after all, over 15 or 20 years old, and new technologies are being developed, such as the 5G, which is very high performing. Manufacturers are very embarrassed because they are stuck having to integrate into their vehicles DSRC technology that is becoming a bit outdated, when new wireless 5G technologies are arriving on the market and would be more exciting to implement. We’ll have to see what happens. We don’t know yet what decisions will be made or which technology will dominate.
Senator Griffin: In terms of your slide that has CAV and uniqueness for Canada, I have one item you might want to add as a bullet, and that is rural and remote areas. I mean, that’s an issue for Canada, obviously.
I come from Prince Edward Island. We have the Confederation Bridge between the Island and New Brunswick, and at the other end, we have a ferry boat service between Prince Edward Island and Nova Scotia. I’m trying to visualize how we’ll connect it and how automated vehicles affect those two types of transportation.
Mr. Qiu: Automated vehicles is just one general interpretation. Should it be called automated transportation? It would include different modes, the vehicle and sometimes a ferry or boat, as it can be, as part of the automated object.
Senator Griffin: So the ferry will also be part of the automation system? What about the bridge?
Mr. Qiu: The bridge, I think, will be part of the infrastructure. We can make the bridge more intelligent and smart, and even the bridge can talk with the ferry and the boat, and boat also can talk with the bridge. In the future, we will consider that. Transportation is like a system, and pedestrians, bicycles, vehicles and boats all will be part of the whole system in the future.
Senator Griffin: Okay. It should be interesting.
Senator Cormier: Thank you for your presentations. You clearly understood that we are aware of how complex the issues surrounding the design of automated vehicles are.
My question is related to that of Senator Bovey, no doubt because we both come from the arts and culture community. We are trying to find quick solutions to complex problems or vice versa, perhaps.
I would like to come back to the following scenario. If, tomorrow morning, you found yourself responsible for implementing the first recommendation that you made of developing a comprehensive strategy, what is the first thing that you would do and who would be your initial partners? Most importantly, how could the federal government, the one that would be giving you this mandate, help you to implement this strategy in the short term?
Mr. Gingras: What would my salary be?
Senator Cormier: That depends on what you propose.
Mr. Gingras: In all seriousness, I think that the first step would be to determine the most important players and establish a round table in order to develop a strategic plan. Several steps are required. I think that one of the niches, particularly in Canada, is test validation. I spoke earlier about simulations, virtualization and all of the IT aspects, but we also need to consider the legislation, promotion and economic development aspects, as well as how to go about launching the first stages of a pilot project.
You may have heard about the idea of making a sort of triumvirate in the Montreal region made up of Longueuil, Montreal and Blainville in order to get these vehicles running smoothly in the urban area of Montreal and test and validate the interaction between the vehicles and road infrastructure in Longueuil, with the help of instrumented infrastructure. Of course, Blainville is home to test tracks, but technological building blocks would have to be added to the site so that intelligent, communicating, and autonomous vehicles could be validated and tested there. Unfortunately, right now, the site in Blainville does not have all of the necessary IT, parallel computers, simulators, and so on. That will cost a lot of money, and I think it is urgent to upgrade the Transport Canada site and give it the technology it needs to become competitive on the international market.
There is a lot of work to be done. I spoke about education earlier. A strategic plan would have many different components. For each component, we would need to set up subcommittees made up of experts from the provinces, the federal government and various agencies that would make recommendations. This is not a two-week project. It is not something that can be done in the very short term, but I think that that the market penetration rate for autonomous and communicating vehicles will be a lot slower than people think. In that regard, my opinion differs somewhat from that of my friend Barrie Kirk from CAVCOE. Many people think that, by 2020, the market will be flooded with autonomous vehicles. I don’t think that will be the case at all. It’s a bit like the DSRC technology. In the 1990s, everyone thought that we would have communicating vehicles within two years. Twenty years later, we are still at the pilot project stage.
We need to take the time to do things right, but we have to take action. I agree with all the other witnesses who spoke before me that the technology is coming. It’s a train that’s coming and it’s up to us to get on, otherwise it will pass us by and we will have missed a huge opportunity to develop our economy, our expertise, and our knowledge here in Canada. I believe that this is truly an extraordinary opportunity in every respect in terms of the benefits it will have for society. I haven’t even mentioned the amount of time that is collectively wasted every day in traffic in big cities such as Toronto, Vancouver and Montreal. The social and economic losses we experience every day because of this problem are alarming, but very little is being invested to try to find solutions.
This is urgent. There is a lot of talk about health, national security, the fight against terrorism and so on. However, in comparison, how many deaths are caused by terrorism each year in Canada? Two? Three? Approximately 2,000 people are killed every year on our roads. I think we need to put things into perspective. In order to save lives in Canada, our priority should be to try to reduce the number of accidents and find solutions to improve our mobility system.
Mr. Qiu: I agree. It is very urgent for us to collect Canadian local evidence and to show how connected and automated vehicles can help to improve the efficiency of using our infrastructure. As I mentioned before, it is very easy for politicians to put money for LRT and transit, but I think for any new technology investment, it’s very limited now.
In Canada, we have the global expertise for artificial intelligence. We have got enough respect in the past one or two years, but how will Canadian technology be applied into connected automated vehicles to increase the opportunity for Canadian industry in the future for the global market? We need to have not just one pilot, but maybe we have three, in the west, east and middle, to implement this.
In the City of Edmonton, when I talk with the mayor, Don Iveson, so now to him in order to say, “Okay, Tony, try to come with me.” No. I want to implement it now. We need to have this champion who really wants to try it in the early stage. Any time, for example, other municipalities and other provinces, and then to really see that, I think, that would save a lot of time and give us a lot of new opportunities as well.
Senator Cormier: As Senator Griffin said, these pilot projects are very important in urban areas, but we hope that pilot projects will also be carried out in rural regions so that people across the country will have access to this technology.
Senator Saint-Germain: My question is for both of you and it has to do with funding.
Could you tell me whether research and development is mainly funded by the industry right now? If the funding is public funding, or in other words, if it is coming from government, regardless of what level, what financial dealings would allow for a return on the public investment for governments and taxpayers?
Mr. Gingras: Thank you for your very relevant question. Obviously, money is the sinews of war. I am a strong believer in win-win scenarios. If we have people at the table who are motivated to find solutions to complex problems that go beyond their local jurisdictions and everyone does their part, then everyone will benefit from the positive impacts and everyone will win.
We need to try to develop a more long-term vision and produce outcomes that will go beyond a single political mandate of three or four years. Most investments take time. We need time to find solutions to this complex issue. I believe it is important for the industry to contribute. Right now, this depends on each and everyone’s roles and contributions. There are several options, but I think that it is important that everyone contributes. So, if the government — federal, provincial or municipal — contributes, then the companies that are participating in the project will be able to develop a wide range of products and skills that they will eventually be able to export. We also need to think about the possibility of exporting our expertise and skills, while of course making good use of them here at home. I therefore think that it is really important to come up with this type of solution for taxpayers, of which I am one.
Mr. Qiu: I think it’s assimilated into two phases, and one is for a pilot, and for my project, the first is actually from the federal government. Infrastructure Canada sponsored $1.3 million for the pilot. Just to do the design and the plan and then install in the infrastructure, install on the public road, and then try to test. That came from 2014, and so over three years. The federal government and Infrastructure Canada, more like a CIDA fund, and then an in-kind contribution.
We will finish the project before the end of September this year. Actually, we are finished already, and for the City of Edmonton and implementation, and for the city they would like to put in cash.
So it’s more like the first phase, just to try to test, over three years, and they consider integrating connected vehicles into their operation. For example, for transit, for fire vehicles, and also for the winter maintenance vehicle, and also now, like with GM and put it in their new car. I think this is good training that we have municipalities or provincial governments that would like to use their capital fund to invest.
That second phase is for operation. For operation, as I mentioned, for a connected vehicle, you want it to be an integrated and open and data-driven platform. This is a way we can make full use of the investment. And we can use the same intersection, we just invest once, but it can be used to connect with different vehicles — transit, fire — so to government, this is a way to save the money.
These are the ways we really need to reduce the duplication. So we don’t want to see in our transit, we have one box for operation, one box for network implementation, one box and then provide information to the passenger. We need an integrated solution. So a connected vehicle, you can see that is the way which you can do the integration.
In terms of automated vehicles — and I talked with the City of Edmonton quite a few times, and the City of Edmonton city council, we had one meeting in March this year, and they approved the work with the University of Alberta, and we want to do the pilot for the automated chateau in Edmonton.
So it’s a way for them to think they have a lot of money, millions of dollars for LRT, but they said, okay, so now, if we consider for the 10 years, 20 years, if we have automated technology already, how that will impact the current planning for LRT investment.
We need to have a champion to think in a proactive way for this stuff.
Senator Galvez: I’m thinking and reflecting on this problem of integration in parallel with the work we do here at the Senate. You see, this is a disciplinary problem or challenge. Our bills are in infrastructure, communication, automotive industry and safety and security. So I don’t think in our time we are going to be able to blend everything together, but maybe there is a priority or an order that we must follow. So if we tackle policy and bills, what do you think should be the order that we can follow in order to follow you, the industry?
Mr. Gingras: To answer your question, I see two or three priorities. First, Canada must be proactive. In other words, we must become actively involved in the development of standards. We must therefore maintain a strong presence with our DOT and NHTSA colleagues in Washington. Transport Canada has already done so in the past, for example, with Intellidrive. Barry Pikilis, a good friend of mine, appeared before you some time ago in that regard. It is very important for the federal government to be involved in the development of standards.
My second point has to do with pilot projects. I think pilot projects are a good start, particularly when a community is willing to try something new. For example, right now, the Mayor of Montreal, Denis Coderre, wants to make Montreal one of the smartest cities in the world. He has a strategic plan. Recently, the city conducted trials of the Navya and Keolis autonomous shuttles. These companies work in this area, and I think this is something that could be worthwhile. I believe it presents an opportunity for the federal, provincial and municipal governments, as well as private corporations, to work together to make these pilot projects a success. Pilot projects can help us to gain knowledge and experience and further develop our expertise so that we can then look into introducing legislation and rolling the project out on a larger scale.
In the short term, perhaps we should follow the example of Ontario and quickly plan to develop federal and provincial legislation to authorize pilot projects involving connected and automated vehicles. For example, Quebec still does not have any legislation in that regard. Every time anyone wants to conduct a small pilot project, they have to request the authorization of the Quebec Department of Transport. That is just red tape. A short-term priority would be to eliminate obstacles and promote the development of pilot projects and standards. Those are the two things that we should work on in the short term.
Senator Eggleton: The Society of Automotive Engineers has provided us with levels of vehicle automation. You are probably familiar with these things. We are probably now, in current day, at a driver assistance level one where there are systems that do assist us in driving. We are probably heading into a level two, which is a partial automation dealing with automated systems performing both steering and acceleration and deceleration. That would seem to be feasible anytime, I would think.
Level three is perhaps a big leap. It’s into conditional automation. There is still a driver that can respond appropriately if necessary. I don’t know how far down the line you think it is before we have these level three vehicles on the road. It strikes me as being similar to the cockpit of a big commercial airline where you spend most of your time on autopilot, but you have humans there, just in case.
Then you get into level four and level five. I suppose it’s arguable that by the time you get high automation or full automation, that you don’t have a driver, you have a passenger all the time. But it would strike me that certainly in the transition phases, which could take several decades before you get to anything like level four or level five, one thing the government should do is require that somebody be in control of the car, or capable of taking control of the car, if there are some problems. So how much farther down the line is it before you see that kind of a change, where you don’t really need a driver and the driver becomes a passenger?
Mr. Gingras: Thank you for this very interesting question. I differ a little bit to the actual state of the automation that we have reached now. If we look at some of the models from Volvo or Mercedes or Tesla, with their autopilots, we are closer to two and a half, almost three right now. Anti-collision systems or automated emergency braking systems already exist. They are still not perfect, but they are commercialized and they are available on several models from different OEMs.
That being said, I agree that the big leap is the transition between whether the driver is responsible and whether there is no driver at all. If I want to be honest and be an independent speaker here, I think the OEMs have absolutely no rush to sell cars that would not require any driver at all. That means removing the steering wheel and removing the pedals, a little bit like what Google is trying to do.
The reason for that is, first of all, their business model. They are there to make a profit and to sell cars. If they want to sell cars, they don’t want to change the business model. If they don’t want to change the business model, they want to keep the driver behind the wheel.
Second, it’s a responsibility problem. If they remove the driver out of the equation, it means that they take the responsibility, a little bit like the Volvo statement . If our car is in the autopilot mode, then we take full responsibility for the accident. We cover everything, no matter what happens. Whether the car is responsible or not, we take care of it. They take the risk because they are confident in their system. But nevertheless, most of the OEMs are not willing to do that right now. It’s a huge problem. It can bring a lot of recalls and a lot of insurance cost.
Virtually, the OEMs would become an insurance company as well if there is no driver at all and if the responsibility is transferred back to the constructor, to the OEM. So they are not really in a rush to do that because most of the money they are making right now is by selling units, by selling cars. As long as the responsibility is on the driver’s shoulders, then it is much less of a problem for them. It’s much cheaper.
This being said, I think that there are other factors. The level three is being used right now with Uber, for example, in Pittsburgh or in Singapore, Phoenix, whatever. In their fleet of vehicles, they are being used. Most of the time, behind the wheel, they have some engineers. They are watching the car as it drives in autopilot mode, and they are gathering data in instrumented vehicles. Then these data are being transferred to a huge database. This is a field operational test mode.
Recently, we got access to statistics from the Pittsburgh experience. Uber is driving between 10,000 and 20,000 kilometres a week with its fleet of vehicles. Interestingly enough is that on average, statistically for the last year and a half of driving, they have gathered millions of kilometres. On average, the human must intervene and take back control of the car every kilometre. So we’re still very far from an autonomous vehicle.
Senator Eggleton: Let me ask you something picking up on that. You’re saying OEM, original equipment manufacturers. You’re talking about the traditional auto industry. Tesla may be the new outlier in that, but we now have Google. We now have Uber. We now have all sorts of interested parties getting involved in this.
They don’t have the usual, traditional business case. They’ve got another goal in mind, and that’s to mine the data — tonnes of data. The general thought is that, in future, you will make more money off the data, and you would be better off providing a service as opposed to the traditional manufacture and sale of cars. Can you throw that into the equation?
Mr. Gingras: I agree with you on that. It's just that it will take time. These guys are not building cars, and the volume of cars they are producing is still tiny. There are alliances between those guys and traditional OEMs. Things will evolve.
Is it going to be Volvo that will be dominant with an alliance with Uber or Google and another OEM? I still believe on the short run — let’s say the next 10 to 15 years — that if any of these companies want to massively deploy vehicles, they will have some sort of alliance between traditional auto makers. It could be tier 1 companies like Magna, for example, which is producing vehicles. It depends. I don’t have a crystal ball here, so things can evolve in different ways. It’s difficult to predict five or ten years from now.
Senator Eggleton: But you will still need a person at the wheel for a long time.
Mr. Gingras: Yes.
Mr. Qiu: Maybe I have a different opinion. In the automated vehicles domain, we have two kinds of groupings. One group is traditional vehicle manufacturing, and the other group is the IT giants like Uber or Google. Tesla is between the two.
Google and Uber are trying to build a new business model to operate the automation as a service. This is like the traditional thinking of Internet thinking. Traditional vehicle manufacturers are trying to sell more and more cars. This is a traditional business.
In recent years, you will find a very interesting thing in that we have quite a few vehicle manufacturers that are trying automated vehicles. I talk with two vehicle manufacturers. They worry about over time, they will sell fewer and fewer vehicles. As vehicle manufacturers, if they sell fewer vehicles, that is a big problem to them. So now they will try to work with IT giants to make automated vehicle to offer vehicles as services.
Technology is so difficult to predict. Just 20 years ago, we didn’t know how IT and Internet would change our lives. What’s coming down the road in five years? We don’t know. Maybe transportation as a service will come soon, in the way that Uber changed our style to use taxis.
As a federal government or as a researcher, we need to think in a proactive way on how we can work in a proactive way to prepare better for the future rather than doing it in a passive way. Maybe later on, we will pay a lot.
The Chair: Thank you, Professor Gingras and Professor Qiu, for your cooperation today. We might take you up on your invitation, Professor Qiu. You hosted us last time; we could probably do it again. Speaking of which, we had to delay the trip to Waterloo. We will be doing it in September — difficulties of participation. That being said, university officials from Waterloo will be before us as witnesses on May 30, so we will be getting information.
Tomorrow night, we will hear from representatives of the Automotive Industries Association of Canada and the Automotive Parts Manufacturers Association.
(The committee adjourned.)