Commercial aircraft has been flying with auto-pilot and autonomous approach and landing software for a while, but they don't have to deal with pedestrians darting out in the roadway, the driver in the next lane who spilled his coffee and is swerving into you, or the person in the car behind you texting who does not see the traffic slowing down
Conversely, I'd say these are perfect examples of what manual driving is poor at, and what autonomous driving would solve. You have listed scenarios in which the driver of the other car would be far better off NOT driving, and the driver in the threatened car will most likely not have the situational awareness to react correctly.
Autonomous vehicles, with sensors all around, with communications to adjacent vehicles, and with the roadway, with algorithms that don't have to drink hot coffee and worry about missing the cup holder, and that don't have to respond to that text just this second, and that aren't distracted by the kids kicking up a storm in the back seat, have a far better chance of operating safely.
So, I would have approached the topic the other way around. I would have said, significant as autonomous landing algorithms can be to improve aviation safety, where pilots are highly skilled at their job, imagine how much more desperately we need such technology in cars.
@Bert: And not just that, think about how software driven cars will efficiently be able to churn out the most power without sacrificing running miles. A complete grid of self driving cars will create little traffic jams and thus also increase gas miles.
One positive things , I can say about the driverless cars is the absense of "Road Rage", which almost 99% manual drivers fall prey to.
When a car overtakes you by cutting a lane, when a car suddenly slows down , when a car behind honks unnecessarily and all such incidents cause "Road rage" by which the drivers loose their cool and try to outsmart the other driver thereby creating a danger for themselves as well as the cars around them.
@Prabhakar: You raise a nice idea here. It surely will eliminate road rage, but for that to happen, the changes to be made should be carried out together. Since most of these would work on software, we can assume that the developers would want a safe driving car, and hence this software would regulate speeds, but can it interpret different speed lanes without GPS? Also, until the changes are made, if one out of 10 people have a self driving car (without the ability to interpret speed lanes) then we will be seeing road rages all over, because the software will enable the car to move at a constant speed only, and the drivers behind this car wont be able to go faster on a 60 mph lane, when the driver-less car is moving at 35mph.
From what I know, Google has nearly surpised the engineering challenges with driverless cars. In fact, their latest vehicle has a better accident rating that the average driver -- I forgot the actual stat but it was impressive.
The real challenges lie in cost (we all know this) and LIABILITY. Try sending a bunch of Google software based driverless cars around San Francisco. They would be the hottest liability targets on the planet! Today, you get hit by a car -- sue the driver. Tomorrow, get hit by a car -- sue Google!
There are sadly thousands of people who would literally 'throw themselves' at this chance. Cheers to centralizing accident liability!
@justicej--this is certainly a serious issue that needs to be dealt with. I liken it to spammers and people who hack and send viruses via e-mail. If we know this ahead of time, would we have not implemented the Internet?
We must employ measures to prevent dishonest people from hindering progress. Just my opinion.
We've speculated in EETimes over the past year about the day that self driving cars become so different from mainstyream cars that a separate driver's license is required. Indeed that day has arrived: "A Learner's Permit for Self Driving Cars: Autonomous Vehicles are a reality and state regulators are racing to keep pace" Fortune, November 17, 2014, page 50. I predict that this state of regulatory oversight will remain in place for a long time until finally autonomous cars are considered to be at least as safe as human drivers under the full range of driving conditions. The question remains open about when (or whether) drivers will be required to be trained in the use of the wide range of safety features (radar cruise control, lane change warning, tire pressure, traction control) being incrementally added which may or may not be intuitive in their implementation.