Sorry to be so long to reply to this...been out for a week, but REALLY? And you have designed or overseen the design of so many control systems that your absolutely confident that no control system could foresee the best place to be and hit the gas rather than the brake.
Yes, today's systems that ONLY have access to the brake would have hit the brake as an only resort. But when you carry the discussion to a self driving car, there are no rules established yet, it is quite possible a system could evaluate the situation and determine, FASTER THAN YOU, that the best place to be is in front of the obstacle!
Great point, bpaddock. Your story illustrates one of the big challenges facing autonomous vehicles. As you point out, no autonomous system would have made the decision that you correctly made. But in a world of all autonomous cars, the situation never would have happened in the first place because an autonomous car wouldn't have pulled in front of you under those circumstances. That's why many experts believe that the biggest obstacle to autonomy is in getting autonomous cars to understand the weird decisions that human drivers sometimes make. As one university professor told us, "Autonomus cars are much more predictable than humans."
Back in the 1970s when seatbelts were becoming mandatory, I accepted a ride with a smoking, recently divorced, mom who claimed seatbelts caused people to die in car fires. A familar pattern, every safety advance has come with stories about how the safety system 'would have killed me.' But automating a car is not required as we already have enough inattentive, cell-phone and texting drivers on the road.
An automated autonomous system can watch 360 degrees around the car including from the bumper edges. They don't get intoxicated or run sleep deprived. They don't replay arguments or other distractive day-dreams. Furthermore, passengers don't take the automated systems from the task of driving.
Congress passed laws mandating backup safety systems and audio alarms for hybrids. But neither law can hold a candle to the never blinking, automated systems. We have lots of laws against crime yet we still have crime. But put technology on the problem, something as simple as a door lock, and theft doesn't go away, it just become less common.
I think "Smart Cars" are about to get to smart. I once almost had a accident on Interstate 80, in a construction zone.
Somehow a car that was over packed pulled from between some construction equipment, in front of me.
This driver could not see out any window but right in front of him, and he was pulling **across** the interstate traffic, not going with the flow. He could not see out the passenger window, that was facing me.
He shot out from between the construction equipment about twenty feet in front of me, while I was doing 45 MPH, remember it was a construction zone (and this was not construciton equipment, it was stationwagon full of junk).
The correct solutions to the problem was to floor the gas, so that I could get in front of him while there was still space, and get off on the right hand brim of the road.
Any automated system would have applied the brakes guaranteeing a collision between us.
Should automation ever be allowed to override the judgment of the driver?
Any automation system that I can think of would have concluded that I made a mistake in hitting the gas, rather than the brake.
I would have to agree with the concerns about drivers become less cautious as these systems come into play. When the inevitable comes and the machines do take over for us, then we can truly sitback and do as we like, but until then, the driver should still remain as and act as the one responsible for the actions of the vehicle...much as the pilots today are (even though today we have aircraft and systems perfectly capable of take off, flight and landing without intervention).
It should be the responsiblility of the designers of these systems to include such responsiblity. For example, in the lane wandering, if the car has to take over then it should also bring the car to a safe halt and not continue on down the road. Or in the case of the Caddy, where the intention is to continue on down the road, it should periodically require some active input from the driver, maybe in conjuction with a Heads-Up-Display to assure the driver is still focused on what's ahead. I don't know for sure, but the part on the braking makes it sound like this may already be in place, it applies some braking but doesn't completely take over until the crash is imminent, meaning the result of not paying attention will be a low speed or soft crash with you, the driver, being responsible.
These systems cannot account for everything and until they can do better then you or I, they shouldn't be expected to account for everything, they are specific to what they do and only what they do. Don't include cases where you and I couldn't possibly do any better as a reason why these are faulty systems. Reviews of crashes have shown that in the event a deer crosses in front of you, your best bet is to hit the brakes and run into the thing, don't try to swerve to miss it. The problem with swerving to miss a sudden obstruction is that we are more likely to lose control and do more damage to ourselves and others than the damage and injury of hitting the object. Yet put in the position, how many of us would still swerve even though we KNOW that is not the thing to do. A robotic braking and manuevering system has no "fear" reactions and would be able to make a thorough decision of the best course of action and act accordingly.
BillyMoore, I would attribute the problems that you mention far more to driver inattention and just plain poor judgement, (aka stupidity), rather than drivers becoming more agressive. They don't understy\and thgat ABS just means that you slide straight into the wall instead of spinning into it.
I'm not convinced that these "improvements" will result in better safety in the short term. I can agree that a robotic system would likely outperform (safety-wise) many of the drivers that cause most of the accidents. However, prior to turning complete control over to robotic systems, these systems will all encourage LESS driver attention to the task at hand. The result will be drivers that will devote less and less attention to driving. It's called risk compensation, and it occurrs when something is made more safe, and the operators increase their risky behavior to compensate. Humans will operate to a certain level of perceived risk, and when the activity becomes "too safe to fail", humans will compensate with more risky behavior. When they get behind the wheel of vehicles not so equiped with these great features, the roads will be even more dangerous. I already see this in the behavior of drivers in cars with ABS, traction control, air bags, 5 star crash ratings, etc. Drivers have become more aggressive because their vehicles have allowed them to while keeping the risk to themselves the same. However, the risk to those outside their vehicles has increased. I'm a cyclist and motorcyclist, and I'm considering giving up both due to the increase in driving aggressiveness I've seen over the last decade or so. I agree that any dangerous activity needs to be made as safe as possible, but not at the expense of everyone else.
Yes, fly by wire systems are commonplace in planes today. Their cost is high since they must meet safety-critical requirements. When the plane costs tens if not 100's of millions, a few million for an adequate robotic control system is reasonable.
I would think a robotic control system's tasks are much simpler in a plane than it would be in an automobile (fewer things to hit, fewer things to do, more predictable surroundings, more predictable scenarios, and in a sense the time scale is longer). The pilot's are also trained and paid (and for the most part expected) to be at constant attention so if anything does go wrong, they can take over from the robot. Which is why, when that transfer doesn't go well (as in the Airbus flight from South America to France a few years back) so many people die (even with well trained personnel and expensive, fairly safe, systems).
But automobiles are commodity items. Every dollar counts, so cost cutting down to the bone WILL be done (and aggresively so). This means one of two outcomes. Either normal people will not be able to afford to drive these very expensive (but safe) cars (or as mattd pointed out, wont be able to afford the very expensive insurance on non-automated cars). Or, the government will figure the cost-benefit ratio of what's good enough and our cars will be automated but only mostly safe (30K deaths a year on the road is OK? But perhaps 50K is not? or perhaps 100k?) The latter is more in line with the fed's track record.
3dRob has quite a few valid points, no doubt. My concern is what happens when an exception arises, which happens to many of us a few times a year. Just think about what would happen when the lane-holding system would see the instance of a damaged center line wandering off to the shoulder. I have seen that twice, it was probably caused by some vehicle moving quickly to the sholder to avoid hitting the line of cars behind the lane painting truck. Rare indeed, but it happens and I have seen the results.
And what would the system do about a large cement bag that suddenly becomes airborn in front of the vehicle when it is lifted up by the wind from a truck in the next lane. Hitting it puts a small skuff in the paint on the leading edge of the hood, but at 55MPH there is no maneuver to escape it when it happens only 50 feet ahead. Anything that the system could do would make things worse, if it was done in heavy traffic, which is when it happened to me, once in 2 years, and once about six years back. And I am sure that quite a few of us have had the wind blow an empty trash can into the road just ahead.
Besides those concerns we need to remember that every included response will be vetted by lawyers to assure that the provider could not be held liable in the event of a lawsuit. That in turn means that all of the logic would make such a vehicle a real impediment to the normal flow of traffic. That fact is inescapable, and should not be ignored.
Are they robots or androids? We're not exactly sure. Each talking, gesturing Geminoid looks exactly like a real individual, starting with their creator, professor Hiroshi Ishiguro of Osaka University in Japan.
For industrial control applications, or even a simple assembly line, that machine can go almost 24/7 without a break. But what happens when the task is a little more complex? That’s where the “smart” machine would come in. The smart machine is one that has some simple (or complex in some cases) processing capability to be able to adapt to changing conditions. Such machines are suited for a host of applications, including automotive, aerospace, defense, medical, computers and electronics, telecommunications, consumer goods, and so on. This discussion will examine what’s possible with smart machines, and what tradeoffs need to be made to implement such a solution.