by Mike –
In connection with research for his upcoming book on autonomous cars, author Wallace Wyss has posed another scenario he envisions could happen as we enter the vaunted age of autonomous cars.
Readers are invited to loose their slings and arrows because, not being an engineer, Wyss admits he may be unaware of some coming technology that would save this situation.
He hopes to see posted reader’s comments in any area: engineering, design philosophy, and even ethics.
Here’s the third scenario – The Four-Legged Critter
Let’s say it’s the year 2020 (only 36 months from now). Autonomous car engineers, after much trial and error at the test track (and many dogs sacrificed), have programmed the software so that the robot reacts to a small animated creature jumping into its path while he is proceeding at 35 mph.
It’s been programmed so the robot gets it now, it doesn’t know what they’re called, but it knows if the animate object has four legs, it has to avoid it. First priority, Plan A, will be to quickly stop in its own lane.
Plan B is to go into the left lane (if it’s clear) or Plan C, the right lane (if it’s clear).
Plan D is to hit the creature if it can’t stop in time.
Now let’s gin it up a bit. Everything is the same except the critter is not a dog or cat or fawn this time but a toddler, crawling on its hands and knees.
What does Mr. Robot do now?
Let us know what you think in the Comments.
February 22, 1959 Edition of Closer Than We Think
Click here to read Autonomous Cars – Question No. 1.
Click here to read Autonomous Cars – Question No. 2.
Regardless of what the 4-legged creature is, if the car can’t stop it can’t stop and if it can’t go into the immediate left or right lane it shall hit the creature UNLESS it has vertical thrusters like the new F-35B and older Auspreys!
I think maybe what he’s getting at is a human being might choose to go into the right or left lane even if the lanes were not clear, rather than hit a toddler. Especially if there is some kind of possible mitigation for going to the left or right lane….like maybe it’s not an oncoming car in the other lanes, it’s a parked vehicle or the back of a van or whatever.
Ken summed up what I would hope–that in Stage 5 where there is no steering wheel or brake pedal, the robot car could have recognition software designed into it sufficient to “prize” human life more than dogs/cats/other animals, and be willing to crash into something on either side (but not head on!) to spare that toddler. Call it the “toddler test” but I want to see videos of that test (with a robot toddler) before I am on board as passenger in the robot autonomous car…
I don’t see a problem.
If it’s an animal it will be the luck of chance since animals move quickly and somewhat randomly and if detected soon enough perhaps the vehicle will be able to stop, otherwise, the program should avoid any damage to the vehicle and others around it. Sorry fella.
In the case of a child, the forward scanners should be able to sense the child in time due to the system following protocol and rules of the road. Children on all four are slow, and the car would be following at the correct distance from the vehicle in front and should detect the child. This assumes a residential area and lower speeds.
Actions taken would be a hard brake (depending on distance and speed), or turn away and accept possible collision.
PS: To anyone concerned, I have a number of the “Closer Than We Think” images as seen at the top of the page if you’re interested in what our future will be…..Lol