A little over a month ago, I was in an accident. It was not as bad as it could have been but was still worse than a little fender bender. It was my fault and several human factors contributed. The first was traffic. Traffic in Silicon Valley is at an all time amount of awful. What’s worst about it is how traffic will pick up then all of a sudden stop. My accident occurred under these circumstances. I was on a freeway I don’t normally drive during heavy traffic times and was not familiar with the heavy spots. Traffic had been a parking lot for several miles, when all of a sudden, it picked up as if there was no problem. I started to accelerate because it seemed as if the traffic issue was over. I had several car lengths of distance but not the recommended three-second distance. Suddenly, traffic came to a (literal) screeching halt. As I went to hit the brake my foot slipped and I hit the gas pedal instead. I hit the car in front of me.
A month before my lease was up on my work commuter car, I traded it in for brand new Prius. I was still adjusting to the new pedals which is why I think I didn’t hit the brake cleanly and my foot slipped. As with many accidents, it happened fast. My airbags deployed and rung my bell. I was going about 25-35 MPH. As I reflected on this experience, I realized just how big of a deal it will be when all cars on the road have sensors that can prevent such situations of human error and avoid accidents and, ultimately, avoid injury.
Many high end cars, like some from BMW and Mercedes, have sensors which will stop the car automatically if it believes it is going to have a collision and the human operator does not react in time. Commercials demonstrate these features using use cases like backing out of a driveway and the car stops before backing into a child passing by on the sidewalk on their bike. Or driving on a dark windy road and stopping before going head on into a fallen tree. But the use cases are seemingly infinite when it comes to accident avoidance due to human operator error, or most likely, lack of reaction time.
These sensors along with assisted breaking and, in some cases, assisted wheel control, have huge potential to make our roads safer. But a high level point to be made is all cars on the road need to have these sensors for maximum safety. Most of the use cases today help cut down on a single human error, primarily by braking for them to avoid collision. But unless all cars have this feature, we can’t avoid the mistakes of multiple humans. Even if my car stops to avoid me hitting the car in front, the car behind me can still hit me unless it also has sensors. Or a car that swerves off the road toward other cars can correct itself while all other cars around it make necessary speed adjustments. If all cars can talk to each other and visualize the world around them to make adjustments for human error, we may have a world free of accidents. Interestingly, this point of all cars needing to have sensors, cameras, CPUS/GPUs with visual processing power like ones being developed from companies like NVIDIA and Qualcomm, is true also of autonomous vehicles.
I was chatting with a friend who happens to be in the automobile industry and is close to autonomous car research at his company, and he was explaining to me one small proof of this. Let’s say, for the sake of argument, a fully autonomous car comes to market and I purchase one. For this car to have passed all regulatory restrictions it will have to have a number of road and traffic safety rules programmed into it. One would be the “three second safe following distance” rule. So I’d be driving down the road, at the speed limit, likely in one of the right lanes and the space front of me would be a three second gap to the car in ahead. As soon as a car (inevitably) pulls into the gap front of me, my car would automatically slow down to give a three-second gap to the new vehicle. You can see how this could cause issues for traffic behind me as well as slow my travel time down significantly. Now if all cars were autonomous, it would be a different story and, in this case, the three second rule would not need to exist. If all cars were autonomous and could talk to each other and make adjustments in real time, cars could travel down the road ver close together. Not only would we have no accidents, we would have no traffic jams. The downside in this is when you stop and think about how long it will take for all cars to have these sensors and all cars to be autonomous. Unfortunately, we are several decades away, at a minimum, from this reality.
Ultimately, the hybrid autonomous concept is likely to happen first. My car will be able to make decisions for me in many circumstances to avoid or minimize accidents. And, as long as there are cars on the road that are 100% controlled by humans, the risk of automotive accidents remain high. Then, in the future, accidents will be a thing of the past.
40 thoughts on “I Believe in a Future with No Car Accidents”
“Unfortunately, we are several decades away, at a minimum, from this reality.”
From what I’ve read, it doesn’t seem like the reality you mention is nearly that far away. What evidence do you have that it is?
Every car sold must have the tech then every car on the road must be replaced. Given length of car ownership cycle, and length of time for auto companies to have standard in every car its a long way away. Folks I know who are insiders to the car industry use stats like 30% of cars will have these features, including self driving / avoiding capabilities by 2030.
And dare we bring up all the non-car thingys on the streets that don’t easily assimilate into this fully autonomous world — motorcyclists, bicyclists, pedestrians, animals, skateboarders, mall cops on Segways chasing skateboarders, etc.
I’m fascinated by the **idea** of driverless, fully autonomous cars but I can’t see it happening anytime soon unless government rams it through…and I don’t see why the government would.
I agree, getting self-driving cars reliable enough is going to be a huge challenge.
A car that can drive itself for 999 seconds in a row without major incidents is quite impressive (although this is only 18 minutes). Now take a car that drives 10,000 miles a year at an average speed of 35 miles; if that car drove safely for 999,999 seconds in a row then it would still have one accident/year. Adding nines is incredibly expensive, because the car would need to be able to deal with some very exceptional circumstances to become this reliable.
As of today, low in my list of concerns is the idea of being killed by a software glitch. Sure it could happen, but the risk is infinitesimally small.
Automobiles relying solely on algorithms and sensors changes that calculation. And the idea of my car requiring an internet connection for software updates and map data is absolutely out of the question.
Downplaying the security, safety, and privacy risks of these sci-fi pipe dreams is maddening. Sure, some driver-assist safety features that don’t require connectivity are inevitable. But the automaker who bets that these autonomous cars are desirable will learn about bankruptcy the hard way.
I don’t think that the ownership cycle is a good input data to predict this “no accidents” future. The birth of the autonomous vehicles world will be such a gigantic revolution that it can’t be reduced to car ownership, especially when the ownership will become irrelevant. Population acceptance and regulatory/legal issues will be the major hurdles.
I agree. People think the obstacle is technological. It’s not. The obstacle is social and societal. Can you imagine American individualism yielding ‘freedom of the roads’ to a horde of robot cars? A lot has to change in the American mindset for that to happen.
Very true. We’ll likely see autonomous cars first in countries founded on collectivism, rather than countries founded on individualism. So probably places like Denmark, Finland, Sweden, those sorts of countries.
“…in countries founded on collectivism…places like Denmark, Finland, Sweden…”
Perhaps “founded on” is the wrong phrase, but certainly collectivism is the dominant culture. Also, don’t confuse collectivism with communism. I get that a lot whenever I use the word collectivism. That’s not what it means.
Perhaps it would be even simpler to say “in counties with good governance, rather than in the phony democracies where the rich make the rules.”
But in any case, wide spread use of self-driving cars is never going to happen.
Also true. America is really a plutocracy. I do think self-driving cars will happen, but it will be especially slow in the US.
You mean it will happen slowly or when it happens, self-driving cars will be traveling the roads at especially slow (average) speeds? I also believe that self-driving cars will happen but the main users will be disabled people who physically are unable to drive and dangerous older drivers who mow down pedestrians like bowling pins. I wonder if the government would allow self-driven cars to be used without an adult or licensed driver in it. i.e. to ferry kids of harried parents to school, extracurriculars, etc.
I mean the US will be one of the last places self-driving cars will be widely used. That’s due to individualism which has also led to America being more a collection of 50 small countries than a cohesive nation. But this also might mean that certain jurisdictions in California could be among the first to implement self-driving cars.
My guess is self-driving cars will be deployed last, if at all, in those places where folks tend to shoot down drones that hover over their property. 🙂
Long live ‘Merica! 🙂
I think that’s a wild exaggeration as applied to the US. I’m sympathetic to the point, but a respect for the rights of the individual is not synonymous with the rich making the rules. I would agree that the two are somewhat conflated by our current politics.
I will stand by my prediction that I made in 2012: It will illegal for humans to drive a car in 2060 in the USA.
What about non-humans?
That would never happen. That would be like saying guns and alcohol would be illegal in the USA by 2060. That would also require the USA spend trillions on a good mass transportation system – which will never happen. Americans won’t stand for their income tax be raised to pay for this.
I stand by my counter-prediction made today. That will never happen.
The average age of the U.S. Car fleet is 11.5 years. Technological progress in the car industry will take quite a while to reach a substantial majority of drivers. In addition, real progress in driver safety will also involve upgrading of the road infrastructure. That said, I think that Ben is right to assume that technological progress will first come in the form of driver-assist systems.
And it will end at that unless governments mandate an abrupt transition date where self-driven cars lose substantial road privileges to make the roads conducive to autonomous cars. Without that, autonomous cars will not be an attractive proposition for most drivers. (More expensive and yet gets you there last?) Which of course will never happen because self-drivers will rise up in revolt.
Actually.. just for the record, we have both fully autonomous rail and air travel. Its regulations that make sure there is a human able to take control if needed.
But planes for instance can take off, fly themselves, and land on their own if we wanted them to. Current tech.
That’s my whole point, we can but nobody wants to. As I said elsewhere, the obstacle is societal not technological.
What have you been smokin, Punk?
Might hybrid autonomous driver-assist systems meet the ‘good enough’ threshold?
It would seem to get us almost all of the way toward the main human benefit — turning auto-related injury/death into a freak accident. It would also likely achieve most of the benefits desired by the insurance industry and law enforcement. The traffic management issue might lessen alongside the growth of a gig economy, telepresence advancements, drone delivery, and better city planning.
So the primary driver (so to speak) of the fully autonomous crowd seems to be the car-centric business owners desiring an end to those pesky employees. This “benefit” of fully autonomous vehicles doesn’t seem like something that’s going to resonate with many drivers. There are other benefits to fully autonomous, but those problems all seem to have much more readily achievable human-employing solution.
As always, I think the key to understanding the issue is in data. The National Police Agency of Japan releases quarterly data on traffic accidents resulting in deaths and there are quite a few interesting nuggets.
One is that a huge number of deaths are related to elderly people (>65 years of age) either driving, walking or riding bicycles. This makes a very strong case for governments to pass law to require sensors in cars when seniors are driving. It’s even possible that seniors will actually drive the adoption of self-driving cars, sensors and stuff, despite the typically long ownership life of cars.
Also, a lot of accidents resulting in death occur when riding bikes or scooters. Scooters in particular are very cheap to own, and are very convenient and common in many asian cities. I’m sure that sensors can help, but I still suspect self-driving cars will have trouble when these vehicles are buzzing around, and we can be sure that self-driving scooters will not become a reality any time soon.
A self-driving scooter could happen if it had a large stabilizing gyro. Aside from that, it would have the same problems as self-driving cars, “intelligent” understanding of traction conditions being perhaps the most difficult. Neither will be practical in the next decade, probably two. Intelligent accident-avoidance technologies, on the other hand, do have great promise in the near future and might help with a lot of the low-hanging fruit you’re referring to.
QUOTE: I traded it in for brand new Prius. I was still adjusting to the new pedals which is why I think I didn’t hit the brake cleanly and my foot slipped.
RESPONSE: YOU HAVE NO RIGHT TO BE ON THE ROAD. This means you are an UNSAFE driver to begin with. You should never be on the road if you CANNOT BRAKE RELIABLY. If anything, you should have practiced, practiced, practiced off the public roads until BRAKING IS SECOND NATURE. This means it would be impossible for your foot to slip off the brake peddle and crash into the car in front of you. Crashing into the car right in front of you is a TEENAGER’S MISTAKE that warrants FAILURE ON THE DRIVER’S EXAM. You could have killed a baby if a baby was in the car in front of you.
There are too many bad drivers on the road who don’t know how to properly drive their cars. Raising the minimum age for driving would help. Suspending driver’s licenses would help. Good mass transportation systems including taxis would help take drivers off the road where they don’t belong.
WOW you really feel you have to CONDEMN OTHER PEOPLE to stop their stupidity AND PUT YOUR WORDS IN ALL CAPS to make sure the idiots GET YOUR POINT.
Good thing we have you to SHOW US THE RIGHT WAY.
I love the dynamic cruise control on my 2016 Audi A6. It makes stop-and-go traffic so much more pleasant. I just set my desired max speed for the road and the car automatically slows and speeds up depending on traffic conditions. Someone cuts in front of me and it automatically adjusts. Now, if it would only recognize a curve and slow down appropriately I would be delighted.
I maintain that the more autonomous or semi-autonomous cars there are on the road, the faster road travel becomes for those people who choose to drive themselves. When you are driving on a road populated mostly by defensive drivers, a little aggressive driving on your part will have all those cars yielding the road to you. Imagine how easy it would be to weave in and out of traffic in the highway if everyone else observes the 3-second safety gap.
Aggressive driving doesn’t gain you much time. It is surprising how much we overestimate the time we think we’re saving by weaving and changing lanes, etc. All you’re really doing is wearing your vehicle down faster and using more fuel than you need to. The actual time savings from aggressive driving is very small, and easily negated by the cost of fuel and vehicle maintenance.
No, not in today’s driving environment. But with autonomous cars out on the road with you? I can’t imagine the chief legal officer of Ford or Toyota declaring he has no problems if the software design department programs some tit-for-tat aggressiveness in its autonomous driving cars. No, autonomous cars will be programmed to be at full defensive driving mode to minimize the automaker’s liability risk. So tailgate an autonomous car or get very close to it from the side and it will yield the lane to you. When approaching an intersection, don’t slow down as quickly or as much and the autonomous car will yield to you.
You may be able to eek out a bit more in the way of time savings, but it won’t be significant. And you’ll still be taking a hit on fuel and vehicle maintenance. Aggressive drivers waaaaay overestimate the time they save.
I wonder if autonomous vehicles could also act like traffic cams, policing and ticketing automatically?
If aggressive drivers don’t realize that being aggressive doesn’t really benefit them significantly today, why would there be a sudden turn to rationality and then a change in behavior just because autonomous cars show up on the road?
As to autonomous cars as traffic enforcement vehicles? Sure, Big Brother will go down really well in the good ole U S of A.
You’re right on both counts. I’m simply pointing out the myth of aggressive driving re: your original comment: “the more autonomous or semi-autonomous cars there are on the road, the faster road travel becomes for those people who choose to drive themselves”.
Aggressive driving is only marginally faster, on an average trip you might save a minute, something like that, there’s a bunch of traffic studies on the subject. The time savings is so small I don’t think we should characterize aggressive driving as “faster”. And there’s the downside of fuel costs and maintenance costs. Of course this all speaks to the larger issue of how poor most drivers are, they aren’t even aware of these truths.
Related to what I’m talking about:
‘Now if all cars were autonomous, it would be a different story and, in this case, the three second rule would not need to exist. If all cars were autonomous and could talk to each other and make adjustments in real time, cars could travel down the road ver close together.’
Are you insane? I mean.. you’re obviously a Smart Guy(tm)… but I reiterate.. ARE YOU INSANE?
The problem with your nefarious plan to cause the worst pileups man has ever known is in having to deal with Unexpected Events. You know… like the terrorist that just tossed a rock off an overpass into oncoming traffic going bumper to bumper @ 65-75 mph. It lands and causes the first car into an uncontrollable state whereupon you get a domino effect that cascades throughout your glorious utopia.
Lets say not a terrorist… what about an earthquake with an overpass slamming into the road. Or a sinkhole opens up. Or a deer darts across the highway. Any number of things can happen.
So NO.. while Google can do fun things on controlled tracks they have 100% control over, having a grid of cars going bumper to bumper at 65-75 mph is something that will *NEVER* happen. Nobody is that insane to not leave room to respond to Unexpected Events such that it doesn’t bind the fate of hundreds of cars(and the people in them) together.