Bringing Safety to Autonomous Driving

We see them on the road every day: drivers who are talking on their phones, texting, eating, shaving, and even changing outfits. These drivers are partly the inspiration for some of the safety technology that is coming into vehicles.

It’s a pretty safe assumption that these types of behaviors behind the wheel will not change anytime soon. That’s why I foresee a future that allows people who do these activities while they’re driving to do them in a much safer way.

For example, here’s a very plausible scenario:

Let’s say a driver knows that he or she is facing a 20-mile freeway commute and traffic is creeping along at a blistering 20 miles per hour. Instead of becoming a distracted driver, the vehicle operator joins a platooning of vehicles, (linked cars in a train-like group following behind one lead car), all headed toward a similar geographical destination and using capabilities like adaptive cruise control. By doing this, regardless of speed variations of the traffic flow, the car maintains a safe and constant distance between the cars in the front and to the sides and ensures that it doesn’t drift from lane to lane.

This actually opens up the possibility of allowing the car to drive itself so the driver can safely attend to those other activities mentioned above.  At Intel, we believe there’s always a safer way to approach these types of situations. We can create a more productive—and fun—driving experience, and do it in a way that isn’t going to put people in harm’s way.

Will the technology to make this scenario real ever be available? Absolutely. It will, and here’s why: automakers are talking more and more about delivering vehicles that are able to drive autonomously. There is a bit of a space race going on right now and the competition will drive innovation to make safe autonomous driving a reality.

Keep in mind, there’s a big difference between an autonomous vehicle and a vehicle that can drive autonomously.  I don’t think car manufacturers are thinking they’ll produce a car that will drive our kids to school each morning after we kiss them goodbye. But I know that a majority of OEMs are thinking about how a driver, if they choose, can turn control of the vehicle over to the on-board system. On autopilot, if you will.

In fact, we’ve heard many OEMs talk about their vision for this technology and even make promises that we’ll see it on the road by 2020. Seven years is not long in the automotive industry, so there’s a lot of engineering, standardization, and policy implementation that needs to be sorted out between now and then to bring this safer way to drive to market. But it will get here, and I predict it will get here even sooner than 2020.

When do you think autonomous cars will be on the market? Do you think this scenario is realistic or not? Why?

9 Responses to Bringing Safety to Autonomous Driving

  1. Som says:

    The “if” of autonomous cars is certain because there is a natural consumer pull for this ability. Recently I saw a man practicing his trombone while driving. The new generation does not see driving as an entertaining experience. Then the ideal transporter is like the one in Star Trek. The next best thing is if you can do what you do in your room but open the door and are then at the destination of your choice – autonomous car!

    Now when this will come to fruition depends on the innovation and risk-taking ability of our generation. When the first practical automobile became available, the UK parliament enacted the Locomotive Act which required that the automobile be operated by a crew of three: the driver, a stoker and a man with a red flag walking at least 60 yards ahead of each vehicle. What regulations will govern the autonomous car? Who will be able to make money on them? That will determine how soon we will see them on roads.

  2. I see two approaches: the Google approach of a car that can handle all tasks independently of the other vehicles on the road, so that the car will be able to take the kids to school, as the above blog suggests is far off in the future. This requires more technology ON the vehicle, but doesn’t require the other motor vehicles to be linked to it. It requires being able to sense the presence of pedestrians and cyclists, even it not in the “view” of the driver’s seat.

    The second approach, ITS (intelligent Transportation System) has a lower ambition: just freeways and only for that part of a trip, not much more than active speed control. It is simpler, but it requires all other vehicles on that road to have the same control and be linked. It is also cheaper, but it will have fewer benefits to society (I am interested in seniors who can’t drive needing to get different places).

    The more expensive approach is more exciting, but the higher cost will both delay its introduction and make carsharing forms of car access more attractive (to spread the fixed costs over more drivers).

  3. Gabriel says:

    This is an interesting topic in this blog indeed. I just can not wait to see this super technology comes into reality, if Jesus tarries. This will b the best for the auto manufacturing. But my worries is, might lead to more people not wanting to buy their own vehicles any more if such exists, and this will invariably turn out to be a disaster to the vehicle manufacturing companies. All the same, I trust that things will surely be as planned if we (Intel inventors) are involved in this.

  4. Chunka Mui says:

    Passenger vehicle platooning might be soon be technically viable but it is hard to imagine how this scenario comes to pass. It suffers the general problem with transportation scenarios that require vehicle-to-vehicle or vehicle-to-infrastructure communication: how do you ever get past the chicken-and-egg starting condition and get to a critical mass of vehicles? Who would buy vehicles with the requisite capabilities before the chances are high that are enough other vehicles on the road “all headed toward a similar geographical destination and using capabilities like adaptive cruise control.” Without strong demand, who would make the vehicles? Without the vehicles, who can buy them? etc. It is easier to imagine platooning happening in trucking, where there the heavily travelled corridors are well known and the benefits more easily quantified. Then the business case is easier, including the value of being the network facilitator. This is the scheme that Peloton Technology is working on.

  5. I think the scenarios described in the article are realistic for a number of reasons. One of the reasons is that the technology to make these scenarios possible are already available or they are near fruition. Another reason is that the number of vehicles on the roads are increasing exponentially in almost every geographies and hence these technologies will reduce the stress level of drivers who will be under increasingly more stress as the traffic on the road increases.

  6. No. This is an unrealistic scenario. A lot is yet to be done and secured before we start putting rush hour traffic in the hands of computers.

    Besides that, why are we spending this kind of money to solve a simpler problem?

  7. I think 2020 is a good estimate. The technology needed is already available but it’ll come down to who can put the pieces together first at a price consumers are willing to pay. It would require an upgrade to the software used in cars with a Bayesian network or adaptive AI learning tool, coupled with more powerful on-board computers and a cost effective LIDAR sensor. However, LIDAR that is capable of performing what is currently asked would put these cars way out of the average driver’s budget.

    I love the idea of adaptive cruise control. As someone who drives a total of 160 miles a day for work, I’d be willing to trade in my car for that kind of tech. It’ll take a lot of testing and reporting to get people to take the plunge, most likely. I think a hamster in a wheel charged with powering and steering a car may be a better driver than some of the people that share the road with me each day… you know, cause of phones.

    • Andrew – we share much in common, parts of a name (Andrew is my middle name, and sometimes my last gets spelled similar to yours), and also a practical point of view. I love the technology that is possible, although we both know that price will make all the difference in adoption. Sensor costs will come down, but only when volumes go up. Automotive will only put a small dent in the volume of sales for LIDAR and other sensors, but the overall IoT could make a change. I’m excited to see standards being pushed that will spread that potential volume across all sorts of industrial applications.

      As it turns out, the next generation “hamster in a wheel” steering the car (a central computer brain) could be the lowest cost component of the equation if it has enough power to eliminate the need for heavy bumpers and body reinforcements.

    • Sam Lamagna says:

      Thanks Andrew.

      I completely agree that affordable pricing is a key to adoption. It will take time for many to extract the value of saving lives, so the ramp must be gradual for the consumer. Everyone loves LIDAR, because it is so amazingly accurate, everything in autonomous has to be perfect, and accuracy is a big part of it. We are seeing really cool things happening such as silicon LIDAR, where there is no spinning parts and potentially high volume manufacturing could make it affordable. But what’s most important you mention is a decision making brain (computer) surrounded by fantastic software. We’d like to be a big part of those developments as we head to 2020