How will self-driving vehicles “drive” us?

During your lifetime, it is very likely you will encounter a self-driving vehicle on the road. But I will not try to convince you into this fact. My intention is to try to steer up some of your thoughts about this emerging technology.

Connected vehicles

Connected vehicles – NHTSA

The development of sensing, communication, and in-vehicle computing technology has enabled the development of new technology – vehicle will be able to detect the surrounding environment, talk to other vehicles or the road infrastructure, and compute all the needed parameters in real time using its own computer. Basically, vehicles will have robot-like capabilities. As a consequence, this technology has capability to eliminate the need for human driver. By eliminating the human error in driving, this technology has capability to introduce significant safety benefits. National Highway Traffic Safety Administration (NHTSA) states that besides improving highway safety, this technology has potential for increasing environmental benefits, and expanding mobility (especially for older people).

In addition, this technology has an enormous commercial potential. Probably, you have heard about the initial developmental efforts started by Google, with the largest project for developing self-driving cars so far. Now, we have a situation when many car manufacturers are jumping in developing self-driving vehicle technology (e.g., General Motors, Audi, BMW, etc.). Besides car manufacturers, there is a long list of other companies (e.g., mobile service providers) that also see their role and potential in influencing the development of this technology.

From this, we can see that government agencies see the potential benefits for general public, and the influence from commercial sector, that perceives economic potential of this technology. Consequently, these two forces will definitely push this technology forward. In addition, as long United States are concerned, there are not a lot of legal constraints preventing these cars to be in the street as of today.

The importance of this technology is clearly evident. And self-driving vehicle is coming (you can see here  how might it look approaching an intersection from inside of a self-driving vehicle). However, considering significance of this technology, everybody seems to forget talking to the third player in the game. You. The user. Besides reducing immediate negative effects of traffic, or boosting the economy, there are other questions that seem to be generally overlooked in the development of self-driving vehicle technology. What I see as a general point for concern, are potential negative effects if this technology is designed primarily with a vision to resolve our current issues. Narrow design vision might result in negative effects in the long term and once this technology is implemented on the large scale.

To find an example of the effects of narrow design vision in transportation, we don’t have to look far. As a very good example, we can look at the conventional traffic signal control technology. This technology (including devices, theory, practice) is still dependent upon the foundational vision, established more than a century ago (for a brief overview of historical development of traffic signal technology, please refer to the previous blog entry). I, as a signal control expert, know the amount of efforts engineering community has dedicated and will still dedicate to improve traffic signal technology. Technology that was originally developed for controlling trains but it was transferred to urban intersections in an attempt to resolve some safety issues. And the consequence is – over and over again, traffic signals have under-performed – never completely resolving crashes or congestion on our streets.

Some folks in Europe have recognized the obsolescence of some of the conventional traffic signal control principles – the same principles that through large-scale implementation and long-term development show to be too narrow and started to create unwanted side effects of its own. Organized in the project called Equality Streets, they are urging for removal of traffic signals and change in design perspective by utilizing natural human tendency for cooperation and altruism.

Now you probably, think – well, what could be the problem with self-driving vehicle technology? First, you might think – well, self-driving cars are highly dependent upon computer technology – and computers are unreliable. Blue screen of death. But then think how often you have seen blue screen error on your latest version of Windows, and compare that to how often you have seen unsafe human driving during your daily commute. So computer reliability is an issue, but not an unsolvable one. Essentially, the deployment of this technology can be delayed until all the possible technology error scenarios have been tested, and several layers of protection have been established. Until it all operates smoothly.

On the other hand, remember that these vehicles will essentially be robots, transporting people around and making probably hundreds of decisions in a minute. Here comes a famous “trolley problem” – what if the car is faced with a choice between doing nothing, where five people die, and taking action, where one person dies? So now you are starting to realize there are ethical issues that these robot-cars will have to face. Consequently, they will essentially need to be moral robots.

But what about moral humans? And what about other effects on humans?

Let me try to expand your perspective here. Imagine you are going to vacation. Your self-driving car is approaching an intersection. Simultaneously, another self-driving car is approaching that same intersection, but from another approach. You have a way to know that the other car is transporting a woman in labor on her way to the hospital. Would you tell your self-driving car to yield to that vehicle? What if that other car is transporting a person that just had a heart attack? And how about if you are late to an important job interview instead of going to a vacation? In these cases, contrary to the trolley problem, no one is bound to die on the spot. But there is a risk of potential harm in the near future. And there is a potential that some other day, you will be the one going to the hospital, and needing to get through the intersection as fast as possible.

So, all delay is not created equal, and that “unwanted” delay might happen to you too. This becomes an issue of access. Every intersection, as a point in the network with limited availability of time-space, often requires the restriction of your fundamental right to freedom of movement. In addition, considering that intersection are public investment, control also relates to the right to equal access to public service. As a consequence of the restriction of these two rights, there is an indirect effect on fulfillment of your other rights: right to life, right to work, right to leisure, right to standard of living adequate for health, right to education. As a consequence, the principle that robot-cars will use to determine their right-of-way through the intersection might affect you and other people on several levels. For example, a person waiting excessively at an intersection on his way to the hospital might die. A person waiting excessively at an intersection might be late for an important job interview, leaving his family without income.

From these points comes the question – should ethical decisions solely be in the responsibility of robots or should humans still share the part of decision-making responsibility? And if so, what will be the effect on human decision-making and attitudes in relation to ethical questions? Would someone, several centuries from now, see as acceptable not to let that woman in labor go through the intersection before him just because his robot car makes decisions based on some principle of global system efficiency? And what if we decide that the principle for obtaining the right of way through the intersection bases on your ability to pay for it? Should some great grandson of Bill Gates (no offense), with the ability to pay more than you, and while going to the beach, get the right of way through the intersection before you? Even if you had a heart attack because you were working on a low-paid job?

These are just some questions to steer up your imagination. At the end of this, we can see that this technology will affect, not just our travel behavior, but our attitudes and values. Consequently, it will affect our development as human race in general. And it might force us to ask some fundamental questions about our civilization.

On the other hand, we engineers often try to address the “how” questions – how can we resolve this? How can this be improved? How can we solve the problem? But we often tend to forget about a more fundamental questions preceding the “how” questions. So this is a message in the bottle to you – the user – to look beyond the potential of self-driving vehicle technology to solve our current issues – traffic accidents, congestion, pollution, etc. Because there is a potential to solve those issue without compromising the potential for greater positive effects on a social scale. Here, I urge you to ask yourself and request from government agencies and tech companies to ask themselves two questions:

  1. What vision of life will self-driving technology have?
  2. What will be specific social effects (on human behavior, attitudes, and values) of this technology – besides economic or environmental effects?

Once we have some opinions on these questions, and you start thinking of some other questions, we can all engage in a discussion and actively shape the design vision of this revolutionary technology.

Until then – imagine…


About Milos Mladenovic

I am currently a PhD student in the Via Department of Civil and Environmental Engineering at Virginia Polytechnic Institute and State University. My focus area is road transportation systems engineering. My primary areas of interest and expertise include Traffic Signal Control and Management, Intelligent Transportation Systems (ITS), and Infrastructure Management Systems for ITS.
This entry was posted in Auto and tagged , , , , , , . Bookmark the permalink.

16 Responses to How will self-driving vehicles “drive” us?

  1. On the computer model I didn’t see pedestrians, cyclists, disabled, approaching and passing through the intersection. How are they influenced by, and how they influence on self-driving vehicles, and eventually on all other traffic flows? Does the new way of traffic flow (self)control is applicable just on intersections with multi-lane legs where vulnerable traffic flow (pedestrians, and other), preferably, would be physically separated by overarching ramps?

    • Milos Mladenovic says:

      The idea of geometric improvements to the intersection is interesting, and should be definitely be considered along with control solutions.
      As of now, the model operates based on the principle of buffer zones, size of the object passing through the intersection, velocity, etc. – basically all the parameters that enable predicting the trajectory of the user along with the protective zone around – either that be a pedestrian or 18-wheel truck. A pedestrian, being slower than vehicles and needing larger buffer zone around her/him would in that case be able to cross the intersection, as long as she/he has a device that can communicate with the surrounding vehicles. As a fail-back option, in the case a pedestrian does not have such a device, we can still keep some type of audio/visual signals for pedestrians/bicyclist present at the intersection.

  2. Blagoje says:

    please reply us your solutions, ideas and concepts concerning this matters:
    – Will the system rely more on internal or external equipment and/or in what ratio?
    – The costs of such equipment?
    – In what degree will self-driving vehicle tend to exclude human factor?
    – Problem of micro-location (next vechicle) and macro-location (GPS)?
    – Problem of locating intersection aproaches?
    – Centralised vs. decentralised system control (cathedral vs bazaar concepts)?
    – Centralised vs. decentralised database?
    – Automatisation of system updates as it is evolving?
    – Constant online mode, and offline capability?
    – Atmosphere conditions resistance?
    – Breakdown resistance and emergency backups?

    • Milos Mladenovic says:

      Oh, I wish I had all the answers. The point I am arguing for is that this type of emerging technology, with a potential for high and long-term impact, should be designed by hearing many voices. Nevertheless, before I give it a shot in answering your questions, I would like to point that this is beyond 2020 technology, so you should consequently take into account other technological developments that will happen from now until then. Moreover, this technology will be developed by many research institutes and companies, so my role as a researcher is still limited one.

      1. The idea is that core technology should be in the vehicle – for sensing, communication, and computing. External, as you called it, or equipment on the road or in some kind of control/information center(s) would be a second level, and the need/reasoning for this technology would need to be justified.
      2. This also relates to point 1. If we are talking about the cost of added equipment onto old vehicles, or higher cost for new vehicles, I think that it will still be in the range around 10% cost of the mid-size new vehicle.
      3. This is why I am trying to ask the questions in the blog. The way I see it, the goal of any technology is to benefit human kind. And it’s design vision should be developed with this in mind. Because this technology might seem to solve issues like traffic accidents, but what if after few generations people forget how to drive? Or can we make the fail-safe system, when the person inside the car needs to take back the control of the vehicle as in video-game consoles – instead of traditional break/throttle/(clutch)? The need for anthropocentric design of this technology is becoming more and more evident as we start to ask more and more questions.
      4. Vehicle will be equipped with LIDAR, GPS, Bluetooth, WiFi, and other similar types of sensing and comm. technology, so it should be able to deal with the issues of determining and making decisions corresponding to its macro or micro location.
      5. Similar to 4., just as you have now codded GPS maps, vehicle will have the information on all the elements of the network.
      6. The way I see it, conventional control principles (traffic signal control) was something you would call Cathedral approach. My argument for control approach is more similar to Bazaar concept, where through cooperative self-organization, you can achieve the emerging positive effects.
      7. I guess this comes to the issues of privacy, and since this is a fundamental human right, I would argue for database design that protects that right.
      8. You will need to click “I agree” button… 🙂 This is an interesting point, and I guess it should be automatic, to prevent from the people that “will install later”…
      9. In offline, my guess is that you are thinking of comm. capabilities. Well, in the case of comm. failure, or lack of infrastructure, the vehicle sensing capability should be enough. Moreover, other vehicles around the vehicle with comm. failure would recognize that vehicle and adjust accordingly.
      10. Not sure what you mean by this one… I guess the error in GPS measurement? There is always differential GPS, in this case.
      11. Fail-back options and system resiliency is something that will need to be rigorously tested before this technology hits the ground.

  3. Blagoje says:

    Aslo, if some kind of “Priority Points” exsist, are they to be bound for vechicle or driver? Maybe in first case some serial number of device can be used, but in another, how will login problem of multiple users to be solved?

    • Milos Mladenovic says:

      My initial thoughts on this question is that priority points, or credits, as I called them, will be related to each user. First thing that comes into my mind is a good old log in with a user name and password.

  4. Donald Katz says:

    One thing that I haven’t seen discussed is automobile ownership. There really won’t be a need for it anymore, except as a status symbol. Why own a vehicle when you can schedule one to come pick you up whenever you need from wherever you are as part of a subscription service.

    • I think that car ownership is irrelevant to this discussion, because it is focused on the technology of controling traffic flows, whether vehicles would be community-owned or private-owned.

    • Milos Mladenovic says:

      Good point. This is just one of the examples of how this technology might give us a change to question many convetional approaches to transportation.
      Vehicle ownership should definitely be a part of the discussion – especially considering that vehicle is usually static for over 90% of the time. But there are other things, as how will this affect the traffic congestion, and consequent reliability of time of departure.
      What else you think could come into play?

  5. Katerina says:

    I would like to compliment the author for the courage he has in undertaking and presenting such a complex issue that starts to create and will create an ocean of ideas and disagreements among humans, institutions and society as a whole.

    You urged us to ask some questions, one what will be specific social effects (on human behavior, attitudes and values) of this technology- besides economic or environmental effects? Well even now, one very present behavior that might create negative consequences along the way of this new technology (self-driving vehicles) is the distracted behaviors. People nowadays don’t have a full time attention attitude while they are driving and they can be easily distracted by their mobile phones or happenings in the surrounding area. With this technology as you wrote at the beginning, human factor might be eliminated but the question is to what degree? Do you think that people, especially young drivers will abuse this advantage of self-driving vehicles for fulfillment of their own desires – relax, put the music on and leave the robot-like machine to drive you around? What about the alcohol abuse? Since humans won’t be driving the car, will alcohol be allowed? If not, why won’t be allowed? What about possible negligence while smoking in self-driving vehicles? What is the social behavior that will be a just and safe foundation for this emerging technology?

    Also, when it comes to humans, I believe that vehicles will be able to communicate with the humans as they will with other vehicles, the road infrastructure and the surrounding area. But what when it comes to animals? In USA, animals on the road might not be quite as often as it is in other countries of Eastern Europe. How this emerging technology will adjust to those issues?

    You also mentioned that this technology will expand the mobility for older people. I think you are right and it will be a huge benefit for the elderly population. But the age limit for driving that today exists; exist regarding the possible safety issues for them as well for the other humans on the roads. So if there is still some need of human presence during the driving process, what happens if an elderly person can not intervene on time? Will the advantage (expanding mobility) as a general happiness for this social group be morally justified as a matter of weighing up advantages and disadvantages over another social group?

    For sure there is a lot to discuss – the human errors can be completely eliminated with this technology but the humans will still be needed in certain situations. The main question here is why humans might still be needed although the vehicles will drive on its own?

    Also, I would like to mention briefly regarding the priority levels. This is just a small piece of this huge picture. I just want to tackle the issue and discuss with you in future. As you wrote – a very intriguing and important question is “Who will make the decision in emergency situations, the vehicle or the human”? And what will be the foundation to call upon for decision making? Let assume that you have a vehicle with a woman in labor and a man that is late for an interview. At a first glance, everyone will assume that priority right should be given to the woman. But what about that guy who might get the job he desperately needs to feed and sustain a family of five? It is not a matter of life or death at this point but it will have effects in near future. Therefore, my question is whether the utilitarian concept should be incorporated in the development of this emerging technology of Rawl’s theory of justice? Or maybe some other social theory of justice and fairness will be applicable to some of the apparent issues? And why the veil of ignorance for the humans might be the only starting point in designing a fair and just system?

    Security will be the last issue I want to tackle. I believe that understanding the liability issues related to these vehicles are considered and discussed by the Government as well as by the automobile industry producers. But to what degree a human safety is considered in the development of these advances? What will be the definition for safe and roadworthy vehicle? Will there be some new principles that these self-driving vehicle models should have? And what about security breach in the IT systems? Don’t you think that this way assassination will be a simple and easy way to practice? What will happen if someone hack the system that controls and monitor autonomous vehicle transportation? And why those possible unfair and disastrous forms of social happenings might be justified for the common good and greater overall welfare? Therefore, my question to all of you is – should they be?

    • Milos Mladenovic says:

      The question of distracted behavior is one of the big issues for this technology. It is directly related to the problem of will this technology still rely on human control and up to what level that will be – if it even does. I sincerely think that these specific questions can be answered from a perspective of long-term effects on changes in human behavior – changes that after some time become embedded and become what we call a culture. But this becomes the question of design technology for greater social purposes. The danger immediately is if a group of engineers would decide upon this, without asking some other scientists (especially social scientist and philosophers) or without hearing the (informed) citizen opinions. Thus, these are some interesting questions, and they should be taken into consideration when developing this technology. This is similarly related to your third point on elderly people, and the risk assessment involved with the need for human driver.

      My personal perspective is that this technology could be capable to detect motion of the side of the road, and predict the movement of the animal towards the road. Moreover, it would definitely be able to detect the animal on the road. This is from the side of technological capabilities. However, your questions are a good starting point for the discussion on ethical considerations of this technology. Should self-driving car break if it detects the deer on the road? And what about if it is a fox? These two animals might have a different impact on the vehicle during collision. So should the car try to save the animal while still protecting the humans inside?

      There are benefits and drawbacks of each approach – aiming to design this technology to completely eliminate human driving vs. aiming to maintain the fail-safe role of the human driver. I think in the short-term, it would be logical to remain with the role of human driver. But in the long-term, the need for a driver would be completely eliminated – from the technical perspective. But the question is should it be eliminated from social perspective?

      Definitely, the impact of this technology requires a consideration from several philosophical perspectives. Moreover, it might even result in development of new technology assessment methods – especially from ethical standpoints.
      Your last point is something that I once said to a friend – traffic engineers will become “anti-hackers”.

      In overall, this would require holistic assessment of all the effects of technology – not necessarily as the technical obstacles we will need to face for developing this technology (unless they are constraint by laws of nature, such as jerk forces that human body can experience comfortably). I think that this assessment should tell us if the overall benefits of this technology are going to secure a better future and “better” human race in that future. So round, and round, we come back to important philosophical points of human life and its goals.

      I just hope that by assessing and developing this technology, our human race will also think more about assessment and development of ourselves.

  6. Miloš Vodogaz says:

    Dear Milos, first of all congrats on this inspirational article with point of view which is far beyond technical approach of solving problems that we engineers have. The main impression on me left an issue about trolley problem and ethical issues, which can be very hard and difficult to resolve. Wether to “do nothing and save two old people, or to take an action and save mother and baby” is, i think, very demanding decision which requires much more time to be reconsidered by a team of respected philosophers than it could be resolved by (more or less) simple computer in a few miliseconds, especially without any of implemented preamble. On the other hand this issue reminded me on one not so long ago movie (2004) called “I robot” where action is settled in year 2035, and where anthropomorphic robots enjoy widespread use as servants for various public services. They are programmed with the Three Laws of Robotics directives:

    First Law: A robot must never harm a human being or, through inaction, allow any harm to come to a human.
    Second Law: A robot must obey the orders given to them by human beings, except where such orders violate the First Law.
    Third Law: A robot must protect its own existence unless this violates the First or Second Laws.

    At first sight these (or somelike of these) law’s implemented in our new future robotic-car’s may be the solution of all ethical problems that can appear in travel process from point A to point B.

    • Milos Mladenovic says:

      Thank you for the constructive feedback.
      Yes, something along those lines (similar to Asimov’s three laws or robotics) would be desirable to exist – as a control decision evaluation mechanism. However, as you mentioned, this will be quite challenging to achieve. But I sincerely hope, not impossible.

  7. Blaine Leonard, P.E. says:

    Keep in mind that, at this point in time, autonomous vehicles and connected vehicles are two, completely separate technologies. As was mentioned, they may merge in the future, but they are entirely different paradigms today. The autonomous vehicle is making decisions based on what it can “see”, through sensors, cameras, and Lidar. They intend to be driverless. A Connected vehicle is making decisions based on what it can “learn” – mostly through shared data from other vehicles and the infrastructure. This technology is much cheaper, at least in today’s economy. It is not intended that connected vehicles will be driverless – they just provide the driver with enhanced information. They may take over a braking function, or something similar, if they know that the driver cannot react.

    As to your ethical decision about saving lives. First, the idea here is to prevent accidents, so there should be few lives lost. Second, the automobile – whether driven by a human or driverless, will have no idea why people are in each car (heart attack, giving birth, etc.), so it cannot make a decision based on that. Third, neither of these technologies assume that there will be no traffic control. Just as is the case today, the vehicle will be expected to follow traffic laws. If an in-labor woman is in a car today, that doesn’t give them the right to run a red light. This will not change. what will change, is that the connected vehicle (not necessarily an autonomous vehicle) will know the nature of the system; it will know traffic patterns and conditions, signal timing, etc., in real time. It will be able to provide educated information about the best route to take. In all likelihood, both the in-labor woman and the man who is late for an interview will arrive in the minimum possible time.

    Security is an issue that that automobile manufacturers and FHWA are working on now for connected vehicles. (It isn’t needed for autonomous vehicles, because they are not relying on any information they don’t generate themselves). Each connected vehicle will have security certificates which will get sent with each message from the vehicle. The same with the infrastructure. Those certificates will change and renew every few minutes, minimizing the chance that they will be tampered with. Will this system be perfect – likely not. But, it will secure a broadly safe system.

    • М Бјелановић says:

      “Second, the automobile – whether driven by a human or driverless, will have no idea why people are in each car (heart attack, giving birth, etc.), so it cannot make a decision based on that.”

      Computer, communications device, cloud infrastructure, and an algorithm (social justice). We all use that every single day. Its just a matter of putting it all together.

      In my view, the “only” thing missing is political will, by making policies that encourage academics, manufacturers and society in general to start implementing this Traffic Control Technology 2.0.

  8. Milos Mladenovic says:

    Of course, safety is definitely one of the reasons why this technology should be developed.

    However, safety can be a very broad concept if you think about it. If you are rushing to the hospital, because you have a heart attack, the delay in traffic can prevent you getting there in time, and have serious consequences to your life.

    The second point is that development of self-driving vehicle is an opportunity for us to rethink the operational principles for traffic control. The reason for this is that conventional control principles were developed upon an already existing device – traffic signal. The operational constraints of traffic signal, as a device, have also constrained the theory. That is why traffic control theory, i.e., operational principles are primarily focusing on aggregate measures – e.g., minimizing total delay.

    This raises a question of – is it just to have an average delay of 1 min for 10 users, if only one user, on the “side street” is experiencing this 1 min of delay. And is it just if that user is on his way to the hospital, during a heart attack. The conventional control does not have a way to take this into account, and does not even consider this in its operation.

    This is an issue of “design inertia” – maintaining design assumptions that are tailored to earlier technology. My main concern is that we might not think broad enough during the development stage for self-driving vehicle technology, and aim at resolving solely our existing problems.

Join the conversation!

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s