"Computers are famously devoid of common sense, and you can think of this pre-mapping as a way to bootstrap some common sense into the car," Urmson says.
Even so, if the world changes between the time the map was assembled and the time the test vehicle drives the route, it can lead to confusion. "There are things that right now are a challenge for us," Urmson says. "For instance, if most of the world stayed the same but the lanes are shifted—so the physical road didn't move but, for whatever reason, the department of transportation decided we should drive a half lane to the left—that would probably confuse the car today."
There are two main components to Google's efforts: "The first is reliability, which means having the car do the things we expect it to do over and over again; and the second is robustness, which is dealing with unusual situations and still being safe," Urmson says.
A large part of the ability to increase the system's reliability and robustness depends on developing new sensors that can see farther and provide a denser data set, according to Urmson. But the ability for a self-driving robot to deal with unexpected or unusual situations that get thrown at it are what make some question the self-driving car's apparent inevitability.
John Leonard, a Massachusetts Institute of Technology mechanical and ocean engineering professor who led that university's team to a fourth-place finish in the 2007 DARPA Urban Challenge, thinks that major technological hurdles in robot perception need to be overcome before self-driving cars can be deployed on a large-scale.
"I have tremendous admiration for my colleagues at Google," Leonard says. "The performance that they have achieved is amazing—for example, their ability to drive at highway speeds. However, because they are building the maps in advance and then having humans pick out stop signs and street lights and crosswalks and so forth, it's very different than turning a robot loose autonomously in the world with very little prior information."
Dealing with the extremes
Much of Leonard's work has been focused on Simultaneous Localization and Mapping (SLAM). Unlike the pre-mapping that Google's system requires, SLAM would allow a vehicle to drive through the world at the same time that it is mapping it. This holy grail of autonomous driving would greatly increase the self-driving car's ability to deal with dynamically changing information—but even SLAM would not be able to solve all problems.
More than 15 years ago the No Hands Across America team drove from Washington, D.C., to San Diego in an autonomous vehicle and made it 98.2 percent of the way without human intervention, Leonard says. What about that last 1.8 percent? "A key challenge even today is dealing with those unexpected moments," he adds. "To try to get to that 100 percent level of performance there's a common-sense reasoning—one of those elusive goals of artificial intelligence—that no amount of pre-mapping is going to prepare you for."
Even Google admits that they have no good way for dealing with these unexpected moments yet, which is why every test vehicle has two backup humans on board to monitor and take over when the car reacts strangely. "Our program is very much a research program at this point, and we haven't really addressed that issue yet," Urmson says. "If we were to take the people out of the cars today, they'd drive pretty well, and you probably wouldn't notice them on the road until something unexpected happened or some element of the [unreliability] of the system appeared."
Subscription Center
Rights & Permissions
See what we're tweeting about
16 Comments
Add Comment"Suppose 10 human-generated fatalities are replaced with five robot-generated fatalities, is that an ethical trade that society wants to make?" that is a very interesting comment. Ethically, it is a reasonable trade-off but people will see it differently. People will see those deaths as additional as though the number of people killed has increased. It is similar with medicines. A medicine may help a large number of people but if a small number of people have bad reactions to the drug, the general public sees those problems as additional issues and not part of the checks and balances. It is as though the drug has only caused the bad reactions and not helped anyone. The anti-vaccine and naturopath crowds are famous for this kind of reasoning and I suspect driverless planes, trains and automobiles will experience the same thing. The general public has no idea of the number of people killed by automobile accidents. But the first time someone is killed in an autonomous car everyone will know and autonomous cars will be credited with one kill not one million lives saved. The real challenge as always is not in making smart cars, but making smart people.
Reply | Report Abuse | Link to thisIt is not just the perception relative safety, but the perception of loss of control. Many people will not want to surrender control of the vehicle to an unseen and uncommunicative force.
Reply | Report Abuse | Link to thisLooking at the tendency of an even slightly overloaded roadway to develop jams and voids in traffic, it is understandable that drivers will be frustrated that a robotic vehicle would occasionally travel under the speed limit with space in front of it, following an algorithm that prevents a jam miles down the road.
Observing that most of the drivers in certain regions speed, it is understandable if drivers would be frustrated with a car that never exceeds the speed limit.
If you are reading the newspaper or something, do you really care if your self-driven car never exceeds the speed limit? It's possible to come up with a fairly accurate time of arrival - would most people really care what the intermediate speeds are?
Reply | Report Abuse | Link to this"Many people will not want to surrender control of the vehicle to an unseen and uncommunicative force." people do it all the time in public transit and airplanes. I'm not sure about the speeding though. Not too many people on buses or trains complain that they are driving the speed limit. When taken together with you comment about control I think it is more likely that people will feel more secure if the vehicle demonstrates that it follows traffic laws to the letter. Remember, people think that when they are driving crazy, they are driving correctly for the conditions, when the other guy does it, it is because he is a maniac. I think most people would panic if the vehicle started driving the way they drive.
Reply | Report Abuse | Link to thisMany people like driving and won't want to hand over control of their cars to a computer. That's unless there's traffic conjestion, they have to drive the same old boring road with a 50km/h limit and speed cams, they are tired after work, they want to go out for a few drinks with their friends, or are already elderly and not as comfortable driving any more.
Reply | Report Abuse | Link to thisYes, you care that your car is doing the 55 mph speed limit, all the human-driven cars are doing a much more reasonable 75 mph, and you are eventually going to get hit in the rear and killed because you are a mobile road block.
Reply | Report Abuse | Link to this"Suppose 10 human-generated fatalities are replaced with five robot-generated fatalities, is that an ethical trade that society wants to make?"
Reply | Report Abuse | Link to thisNo.
The 5 that are saved are going to be the worst drivers among us, while the computers in the cars are going to kill people at random, including the very best of drivers that have no accidents at all, ever. If you are such a driver, why would you ever get into a car that kills occupants from time to time?
I would want much much more than a 50% reduction in traffic injuries and deaths from my computer controlled car. There's no reason that a driver than never falls asleep, never gets bored, never gets distracted, never feels the need to race another vehicle, never feels slighted when another car cuts it off, should not be able to drop the fatality and injury rate by a factor of 10 or 20. All accidents should be the fault of the other vehicle, and the robot driver should be excellent at avoiding even those accidents.
I think that by the time robotic driving is real and commonplace, the reduction in death rate will be more like 999:1, than 2:1
Reply | Report Abuse | Link to this@rally2xs : less coffee
@everyone else : I hope so.
A necessary step to autonomous cars is the integration of these technologies into standard automobiles as collision avoidance systems. If these technologies could successfully avoid accidents in real world situations by either alerting the driver to potential dangers or by taking limited action (such as braking) if the driver fails to respond to the system’s warning, the value of these technologies would be proven.
Reply | Report Abuse | Link to thisLet these technologies save a driver from a potentially devastating accident and the psychological barriers to autonomous control of vehicles will disappear.
I'm concerned with how it will deal with those tire eating potholes that appear without warning, construction, and the line painting orange cones. Not to mention accidents and detours.
Reply | Report Abuse | Link to thisI'll assume it can deal with the Driver's Ed scenario where a ball comes bouncing into the street, followed by the kid chasing it.
An automated driver might have a better shot at avoiding potholes than human drivers since they could have radar as a sensor, as well as possibly being networked with other cars ahead that have already identified the pothole. In fact, the car might have already downloaded the location of all potholes on the intended route as the car is leaving the driveway.
Reply | Report Abuse | Link to thisI think much could be done that is not possible now via computer communications. The challenges will likely be the dynamic things - a tumbleweed is blown into the path of the car - avoiding it could cause a skid and crash, while hitting it might mess up some paint, depending... what to do? That's a tougher decision for a computer, I think.
We surrender the accelerator to speed control.
Reply | Report Abuse | Link to thisSome drivers now let the car park itself; or to tell them if something is too close and institutes autobraking.
This is just one step further.
At my age I can see great value in a car that can call 911 and drive straight to the hospital emergency room. Or one that can detect the Alcohol on your breath and refuse to go into a bar's parking lot if you are about to reach the legal limit. I would feel a lot safer if I knew, when the bars let out, the patron's car would be the designated driver. Even so, I know several people who would try to drive over a cliff, just to prove it could be done. You also have to realize that in an age when hackers have broken into the Pentagon's computers, the system in the self driving car would have to be totally fool proof.
Reply | Report Abuse | Link to thisTechnology help everyone on the world.
Reply | Report Abuse | Link to this<a href="http://tapchipcworld.blogspot.com/2011/08/steve-jobs-resigns-as-ceo-of-apple.html">Google Plus SEO</a>
Reply | Report Abuse | Link to thisI believe you have just answered their pre-mapping problem. It is very rare that you are the first person to drive along a particular stretch of road.
As data comes in, it could be transmitted to subsequent vehicles traveling along the same path, thus creating the map on a continuous basis. Of course, transient data, such as the kid chasing the ball, could be accounted for as a heads up to subsequent vehicles to use extra caution in that area. A crash could also transmit data to vehicles farther away to reroute around it.
In the event of uncertainty, the autopilot would alert the person in the driver's seat and disengage if needed, bringing the vehicle to a safe stop along the side of the road. For example, if map data were unavailable or outdated(data expires after a set number of hours), you could not engage the autopilot, but it would continue to map out the road you are on for subsequent drivers or possibly for your return trip.
Static road signs could be made machine-readable through RFID and signals could be enhanced with transmitters in the long term to help alleviate the need for pre-mapping.
The marvelous use of vehicle activated radar:
Reply | Report Abuse | Link to thisThe vehicle activated radar is device used to calculate the speed of the vehicles which is running beyond the speed of the law permitted. The people who used break rules often on road. This very best device is developed for to indentify the culprits and show them to the law and get punished.