My most important essay to date
Today let me introduce a major new series of essays I have produced on “Robocars” — computer-driven automobiles that can drive people, cargo, and themselves, without aid (or central control) on today’s roads.
It began with the DARPA Grand Challenges convincing us that, if we truly want it, we can have robocars soon. And then they’ll change the world. I’ve been blogging on this topic for some time, and as a result have built up what I hope is a worthwhile work of futurism laying out the consequences of, and path to, a robocar world.
Those consequences, as I have considered them, are astounding.
- It starts with saving a million young lives every year (45,000 in the USA) as well as untold injury in suffering.
- It saves trillions of dollars wasted over congestion, accidents and time spent driving.
- Robocars can solve the battery problem of the electric car, making the electric car attractive and inexpensive. They can do the same for many other alternate fuels, too.
- Electric cars are cheap, simple and efficient once you solve the battery/range problems.
- Switching most urban driving to electric cars, especially ultralight short-trip vehicles means a dramatic reduction in energy demand and pollution.
- It could be enough to wean the USA off of foreign oil, with all the change that entails.
- It means rethinking cities and manufacturing.
- It means the death of old-style mass transit.
All thanks to a Moore’s law driven revolution in machine vision, simple A.I. and navigation sponsored by the desire for cargo transport in war zones. In the way stand engineering problems, liability issues, fear of computers and many other barriers.
At 33,000 words, these essays are approaching book length. You can read them all now, but I will also be introducing them one by one in blog posts for those who want to space them out and make comments. I’ve written so much because I believe that of all short term computer projects available to us, no modest-term project could bring more good to the world than robocars. While certain longer term projects like A.I. and Nanotech will have grander consequences, Robocars are the sweet spot today.
I have also created a new Robocars topic on the blog which collects my old posts, and will mark new ones. You can subscribe to that as a feed if you wish. (I will cease to use the self-driving cars blog tag I was previously using.)
If you like what I’ve said before, this is the big one. You can go to the:
Master Robocar Index (Which is also available via robocars.net.)
or jump to the first article:
You may also find you prefer to be introduced to the concept through a series of stories I have developed depicting a week in the Robocar world. If so, start with the stories, and then proceed to the main essays.
These are essays I want to spread. If you find their message compelling, please tell the world.
Please add another chapter about security; privacy and liberty
As an engineer myself, I find robocars a very cool idea. And as (it seems) one of the few people who drives a lot but doesn't particularly enjoy it often, I would really like to use them. However, I would push against them for the one reason that it gives lots of power to the government and anyone who reads the news knows that there is an increasing trend in serious abuses of that power.
First is a complete lack of privacy - there are now records on everywhere you drive. I have enough of that with Texas toll tags, photos of license plates and I hear more such sensors are coming.
Second is that if a cracker gains access, even just to monitor, he can wreak tremendous damage to lives.
Third, the system will generate marketing data that will be sold by the state and used against us to place billboards and driver-selective advertising. We almost have this today with cell phone tracking re bluetooth and other avenues. John Smith drives by the McDonalds and a sign practically invites him in my name and his favorite dish.
Fourth, in the event of any kind of protest or revolution, the state can shut down cars, giving them total control over transportation that they currently have over air travel and are slowing gaining on train and bus travel. Imagine a Do-Not-Fly list extended to a Do-Not-Drive list, where anyone with trigger names has their car shut off when they get within a mile of the Whitehouse.
Please consider these issues and add a new chapter to your excellent series that addresses them. The governments are NOT going to get more benign and we must not hand them more keys over our lives. As EFF Chairman all of these and more must surely have crossed your mind in writing your essays.
Centralization bad; robocars good
There may be some substance to your worries, especially if cars ever trust other cars to any extent as suggested by parts of the essays as an option for making them work better. Mostly, they seem to be derived from a belief that the system he is describing is somehow centralized. It isn't. The only difference between a robocar and a car today is that no human is needed to drive it. He mentions that some cars already rely on computers a good amount.
What could be worrisome is paying robotaxis, which can trivially record their source/destination, via credit cards, which can easily be linked to people. Also, a robotaxi might target ads based on that information in addition to appearance and clothing, possibly to the point that it could positively identify someone (say, by comparing the face to MySpace or Facebook photos). But these issues are in the future with or without robotaxis.
Self-driving cars do not magically give any group any more extra power than computers as compared to calculators. (In all fairness, doing your accounting with your computer makes identity theft a lot more likely than doing it on a four-function calculator.) As with any product or technology, security and privacy should be considered, but I do not see them as intrinsic concern for robocars.
I have that chapter
Actually there is a chapter on that subject already. It doesn’t speak only to that, but it is the main focus. I do plan over time to expand it to include ways to get around some of these problems.
One tough one is how to arrange anonymous use of robotaxis. I fear that people will want to assign reputations to robotaxi renters and have a way to track them down if they trash them. This will cause people to push only for identified robotaxi use. There are some solutions. For example, the reputation could be pseudonymous. The user could offer the bond of an insurance company who will be responsible if the vehicle is damaged, without revealing who the passenger is. But it will take work to make these systems be the norm. That’s part of why I write this — to get us ready for these problems.
the leading cause
"it the leading cause of death. "
No, do your statistics properly. Everything is the leading category of anything. Say accident kills 60%, disease kills 40%. The we could say the causes of death are 1. disease 2. workplace accident 3. traffic accident. Just state your percentage, don't try to rank categories that you haven't even bothered to elucidate (let alone justify). Because the time has indeed come for quality "robocar" advocacy.
I will try for more precise numbers
But in fact there are some reasonably well established categories in death statistics, and traffic fatalities have long been their own category with lots of history. They are not over 50% of course, so I get your point.
So whatever happened to the DARPA Grand Challenge?
The last DARPA Grand Challenge was held 11/3/07. Where is the one for this year? There was one in 2004 and 2005, and for some reason 2006 was skipped and so it's apparently the case with 2008, but what about 2009 and beyond? Disturbingly, when I go to the webpages of the finalists, most of them are still acting as if the 2007 event were held yesterday. Is that it? Is it done forever? Of course, the point of the challenge was not to design cars to take over the role of human drivers in city/highway traffic, but I got to believe there is still a lot more these cars can be challenged with before they can actually be deployed in the battlefield. I sure hope it comes back in 2009. Two years is a lot of time to make improvements. In addition to getting the researchers to compete and make better cars, it also gets people interested and more comfortable with the idea.
No new challenge yet
As far as I know there is no new challenge announced, but the prior ones have been such a great success that I would be shocked if they didn’t want to do another. Of course, after some time, we move in to the commercial realm (mostly in selling to military at first) and at that point they may not need prizes.
There was an “exhibition” race earlier this year in Southern California. It’s been suggested to me that soon it may be time to hold a non-academic conference on the matter, which would include contests no doubt. (There are a few different academic/robotics conferences that cover these technologies already.)
Great Idea
I've had a similar idea that I describe here http://highscalability.com/scaling-traffic-people-pod-pool-demand-self-d... - Scaling Traffic: People Pod Pool of On Demand Self Driving Robotic Cars who Automatically Refuel from Cheap Solar. Keep up the good work. It will happen sometime.
Why the solar
I am curious why you think cheap solar is an important part of things. Not that it would not be wonderful, and if we get it, it should be put on the grid quickly, but this is an orthogonal issue. Same with some wonderful new battery — great if we get it, but robocars let us have usable electric cars with old, cheap battery tech. The better battery just makes it nicer.
One thing that some people think of that turns out to be a mistake is solar mounted on the vehicle itself. The power from such panels is modest, but worst of all, it would often be wasted. If the car’s battery were near full (as you would want it to be) the solar power would be discarded. If the car was not pointed the right way or was in shade, the power would be lost.
Solar power on the grid can be angled to the sun (even if fixed) and always generating every watt it can. No watts thrown away, all available to be put into cars or other purposes. While a panel on a car could save a bit of battery capacity, I think the waste would outweigh that. (I am not saying you were proposing panels on cars, were you?)
EV availability
"But few will buy the Tango today."
I suspect this is not so much a distrust of the batteries but more an issue of price.
The design is unusual, but I think it is a great solution for the single commuter.
Availability is also another factor, there are a few interesting EVs on the market but are either high cost high performance or just not sold outside their local area, i.e. California USA!
What we need is a cost comparable commuter vehicle that can cover at least 100 miles per charge at a reasonable speed 50 - 60 MPH.
Maybe the sentence should read
"But few CAN buy the Tango today."
Derek
Who killed the electric car?
While there are arguments back and forth, I think there’s a reason that Hybrids are hot shit (and plug in hybrids the next hot shit) while electric cars did not find a market. Everybody who drives electric cars loves driving them. They don’t love recharging them, nor the cost of the batteries. I don’t think it’s just a case of not having a suitable electric car (with today’s battery issues) appear on the market.
The Tesla will sell very well for a car costing more than $100K, but it’s those batteries that make it cost like that.
One aspect of robot cars
One aspect of robot cars that would contribute to their relative safety is that they're more likely to cooperate with one another. It's one thing to set up a program that guides one's own car... but if the cars can communicate with one another, then we'd see better on-road cooperation. For example, if my car had to switch to another lane to make a turnoff, it could send a request to other robot-controlled cars in that lane, which could ease off and allow me the space I'd need.
As for the issue of centralized control, and the guvmint shutting down drivers... well, they can do this now _anyway_, and it requires a lot of messy and occasionally dangerous police work.
Here's an interesting question; what about emergencies? I can see ambulances and police cars requesting "emergency access" which'd nudge other cars out of the way. But would ordinary drivers be able to request such access in rare occasions (like the classic wife-in-labor situation)?
Emergency access
In my view, if the cars pass the “school of fish test” the ambulance could just plow right through the school of cars, and they would part like water if they could.
The only difference is that the ambulance would not get a ticket for pulling this stunt.
In the more distant, HDV-less future, I imagine roads where traffic always flows very well by today’s standards, and an emergency bot can get to you very quickly in normal traffic.
Where to go?
Hi Brad,
Great series -- really got me thinking about the possibilities, and convinced me that these are a better alternative to PRT, which I've been a fan of for a few years.
So, as an interested computer geek, I'm curious about where to go from here. Are there companies that are putting money into this kind of work? Or will it really take a presidential call to action for things to get moving? If the latter, what's the best way to start getting up to speed on some of this -- building my own self-navigating robots? Learning AI?
Appreciate your (or anyone else's) input on these questions.
Best,
Sean
In existing cars first
DARPA hasn’t announced a new challenge yet, but many hope they will. In the meantime a lot of technologies are being incorporated into cars by companies like VW, Audi, Lexus and others that I list at the start of the roadmap page. Companies making the components — the sensors like LIDAR — are active concerns, though I don’t know of a company that is a pure play yet.
Had not seen your articles til today. Also wrote bout robocars
Great set of articles on robotic cars.
I had written about electric powered robotic cars as well.
http://nextbigfuture.com/2008/07/electric-powered-exclusive-robotic-car....
My belief is that the city robocar only zones is the best way to start deploying robotic cars sooner.
The current versions of robotic cars are very close to being up to the job of navigating slower city streets.
I had not realized how poor the efficiency was of the US public transportation system.
This is of course not the case in places like Hong Kong and Japan and some places in europe where rider density
is far higher.
Robocar only zones
The main concern here is that people will object to robocar only zones in the early stages. They will be seen as elitist (only the rich, who can afford the initial, more expensive vehicles could take their cars into the zones.) However, if congestion pricing becomes common first, this could change things a bit. And the zones can’t exist until the robocars are well tested and trusted with other robocars and pedestrians, in which case handling other cars is not a big deal.
Thus I predict the robocar only zones come later, as a means to de-congest core areas in a world that’s about 60% robocar and 40% HDV, with cheap taxis for poorer people available in the central zone.
I can imagine systems to try to keep the pedestrians apart from the robocars but there still need to be intersections and crosswalks and there will still be jaywalkers even with giant barriers.
In areas with fully independent right-of-way you could get there sooner, but it’s expensive. For example a network of small, robocar-wide tunnels might become cheap with new tunnel boring machines, allowing electric robocars.
As for Asian transit — it is more efficient, but it still isn’t as good as an ultralight single person electric, which is essentially an electric tricycle with composite shell.
Autonomous driving vehicle already done in Old Country
My brother at University of Miami in Florida related hearing a story from our mother, which I hadn't heard first hand, about life back in pre WW-II Yugoslavia in a German-speaking village near Belgrade.
A person could go to a party and get drunk as a skunk (think of Die Fledermaus -- partying and drinking and practical jokes represented in that comic opera were very much the cultural norm, at least for Grandpa's generation and friends). When it was time to leave, you simply needed to be lucky enough to have a friend stuff you into your horse-draw carriage -- instead of play a joke on you by leaving you on a park bench in a bat costume like in the opera.
Then all you had to do was give the horse a verbal command "Home!" The horse was inclined to return to its stable of dry straw bedding and fresh hay anyway, and you would be driven home, automatically and safely.
A horse is not nature's brightest creature, but it seems what you are trying to do is invent a computer and accompanying software to be at least as capable as a horse, or at least with regard to image processing, path planning, and obstacle avoidance. And a computer that doesn't leave "presents" on the roadway for someone else to clean up.
Yes! Been thinking along similar lines
I'm glad I found this article, because I have been advocating this for a while, but you have a much bigger audience.
As an engineer, I'm sick of wasting my time driving when I know car itself could do the job much of the time. I even joined a DARPA Urban Challenge team, but it was poorly run so I dropped out.
I'll bet that the time wasted during driving adds up to an enormous amount of lost productivity, and a tiny bit of that value, spent on this project, could go a long ways.
I'd love to have an electric car (because they are better to drive), but frankly, not being a member of the Church of Climate Change, I'd be happy if my SUV could do the job. In fact, I've been think about how to interface to the cruise control...
Keep it up!
BTW, a long term (expensive infrastructure) solution to the battery charging problem is battery exchange. You drive to a service station, a robotic machine extracts your discharged smart battery pack, inserts a fully charged one, and charges your card based on the cost of the transaction (battery age, energy costs, overhead, a little profit). Notice that this uses a current approach (servicestations) - always a good thing if you can do it, and is more efficient (the power distribution capacity increase is focused on a few places rather than every home) and
Not so inexpensive
Battery exchange, which I discuss, is a worthwhile thing, but it has some difficulties. Charging poles can be anywhere at a very low cost. Battery exchange stations are more major affairs. This requires lots of standardization of the batteries, and a good, safe cartridge design that can regularly make a connection over which hundreds of amps will flow.
The batteries would then have to be owned by the battery company, not the car owner, except in fleets. And the robotics for this are doable but would need to be built. So yes, I see it happening, but I see charging poles first.
You don’t have to believe in global warming to want an efficient vehicle. How about just not wanting to pay $23 to go 100 miles in that SUV? And whether you hate CO2 or not, gasoline pollutes in many other ways.
the future is almost here! i mean it this time!
I want my robocar to fly. And it should mix me drinks while we're flying. For me to even consider buying it it must give me shiatsu massages as a default, without me asking. BJs would be nice, too, and maybe a big 'ol steely dan should be standard equipment for the ladies. And I thought of a cool way to help even more with parking congestion. Why not install some sort of anti-gravity AI? That way my robocar could just float over my destination (e.g. a Star Trek convention) and I could just slide down a rope. Better still, all robocars could come with transporters that could zap us that last 50 feet from our hovering cars to our final destination . Better still, why not skip right over the robocar part and just give everyone a personal or family-sized transporter. That'd be awesome! No more traffic, no more accidents ('cept the occasional genetic blending with flies like in the movie). And all those highways could be turned into bike trails! Or open space! Or community gardens!
Yea. The future is gonna be so cool.
I understand skepticism
But watch the videos pointed to in the primary article and you may see why people are viewing this is something more than science fiction.
Wonderful, another step
Wonderful, another step towards making me more subservient to the whim of politicians.
next, you'll want to automate my thermostat so the A.I. can decide the proper temperature for me.
Wait, that one's going to be here first.
You'll take the keys to my "dumb" car when you pry them from my cold, dead hands.
But how distant is this future?
I get a little concerned over the idea that you seem to be working on Step G (or so), when we haven't gotten past Step B yet: The current leader in "practical robotics" is iRobot; yet their 5th-generation Roomba, which is capable of several things beyond what the Discovery (series 4), can now clean multiple rooms in a single "mission"; yet it is no longer possible to use two Roombas in adjacent rooms (which the simpler 4th-generation Discovery could do).
It seems that the 4th-generation used a single "command center" to program everything to the same (remote) clock, including the Virtual Walls; the 5th-generation Roombas each have their own Scheduler, and now use RF to turn the Virtual Walls on and off -- and the Virtual Walls are "keyed" to individual Roombas, so now, when one Roomba finishes its schedule, it turns off the Virtual Wall that had kept the other Roomba in its own playpen.
Apparently, iRobot thought that they'd had the "simple" problem so well contained that they could increase the "complex problem" capacity, and unfortunately, they overlooked something. (It's commonly noted that "software is someone else's idea of how to solve your problem".)
By your own admission, the first robocars are likely to kill people; but, What If (great Sci-Fi question): Steps C, D, and E get us to the point where there are a number of Robocars on the road, and everything seems to work; steps F and G bring us to your "school of fish" robocar-traffic-solution; and then some tiny, overlooked detail -- the "Naw, no one would EVER do THAT with a car" (as, apparently at iRobot, no customer was ever expected to use multiple Roombas on the same floor) -- causes your school of fish to all swerve TOWARD the fire-truck that all the programmers had expected it to AVOID --
-- and then hundreds of people are killed or maimed in one single "oops".
The shuttle explosion killed seven people, and ground the entire space program to a halt; what will a "cluster collision" (I could've used a different term there, hehe) do to any nascent Robocar program? And how can you be SURE you've solved ALL of the problems before you turn two tons of autonomy loose on the streets?
How far away?
I do include a section on that, but it really depends a lot on political will. When I examined the numbers, and found that robocars could save us trillions of dollars, and millions of lives — like curing a major disease — I realized that this is a project society is stupid not to put a lot of effort into, whether it be private commercial funding or military or university lab research. But there is no crystal ball that can predict the time. My message is that this looks doable and well worthy of a major effort; as much as, if not more so than Apollo was.
The robocars won’t go on the road until people feel they are really safe. They may start as whistlecars and deliverbots with no people in them except when going slow. Designers will work hard on that cascade accident scenario you describe. At first, they will do that just by leaving lots of space, as much as humans do, and often more — as much as the road provides. They simply won’t pack so closely, at first, that this could happen.
They also should not all run the same software, though people debate that with me. In fact, they will run multiple versions of software in one car. If any of the systems in the car says, “Hey, that plan is dangerous” it probably won’t be done, or at the very least the other systems will all re-evaluate.
From a social standpoint, replacing human driving which kills 45,000 in the USA per year with robots that kill 500 a year would be a very wise choice, but I understand how the public will probably not think that way.
I should also add ...
... that iRobot seems to have no idea how to solve the abovementioned cooperation problem; their own monolithic mindset keeps them from seeing anything other than the problems that they themselves come up with.
You're missing the point
OK AI Geeks, let's get back to the REAL point, and stop the masturbatory fantasies about robots doing everything.
When somebody else is holding the steering wheel, THEY decide where the car goes. IF IT GOES.
Let's say you put in your request for the botcar to come and ferry your brood to Grandma's house. You get the following response: "You have exceeded your carbon usage credits for this month, we apologize for your lack of deference to Gaia".
Carbon is going to be produced as a result of every human activity for a very looong time, electricity source notwithstanding. Hopefully AGW will soon be discredited, but that's a rant for another website.
This will be another excuse to control human movements.
I don't care if it will save lives. You can't put foam rubber on every sharp edge in the world, and I'd rather people didn't try. If I'm going to die in a car, I want it to be with my hands on the wheel. I would fervently wish for a software glitch or hack that kills a large number of people. That should put this idea where it belongs.
Yes, there are people who should not be allowed to drive. But every time the difficulty level of everyday life drops, you end up with more people who don't get even a small mechanical aptitude.
And there will always be that gross mechanical interface between the AI and the physical world.
Know what vision of the future scares me? People immobile in their homes, unable to do anything but frivolous cerebral pursuits, travel restricted because power generation is deliberately not expanded, and robots fulfilling most needs. Won't happen to me, I'm too old. But my infant son will, and probably won't see anything wrong with it.
BTW, UAV tech is way further along. Why don't you see airliners going first? Oh, that's right, that one's too obvious.
Post new comment