bobsneidar at iotecdigital.com
Fri Mar 30 19:16:48 CEST 2018
We have computers automate these processes, but always with a human ready to intervene. The computer will act based upon the inputs it receives. If the inputs go wrong, you may have an exceptional diagnostic routine running to detect it and act accordingly, but only a human can make a judgement at "runtime" if you will.
The value to using computers in these instances is that computers can react much more quickly, and given all the corrects inputs, and assuming it is functioning correctly, will always make the right "decision" based upon a predetermined set of parameters. A human can be slow. A human can make mistakes. A human can become malicious. A computer cannot. A computer can however, lack all the inputs to make what we would call a right decision.
Take the classic arguement of a person walking a dog. The dog breaks away right in front of an oncoming car with no time to stop. Does the car swerve to avoid the living mammal right in front of it and hit the dog owner instead? How could you program a computer to make that distinction?
If computers are going to drive cars, then it needs to be done all at once. A switch gets thrown and now no one's car will work unless the computer is driving it. Also, I think it would be better if auto driving cars were restricted to places without crossing traffic, like highways/freeways. A car leaving a freeway would have to come to a stop and the driver should have to take control to proceed on streets.
Ideally self driving cars should never be based entirely upon optics. For this sort of thing to be really as safe as can be managed, every car would need to have beacons at the corners and maybe along the sides, and the cars would need to be in communication with all the other cars within a certain distance of it. People however would not accept this because there would be the potential to track people without them knowing it's being done. (Of course that is likely happening now but that's another discussion).
> On Mar 30, 2018, at 09:51 , Mark Waddingham via use-livecode <use-livecode at lists.runrev.com> wrote:
> In terms of computers driving cars then we let computers monitor and interpret sensors for humans in nuclear power stations. We let them 'drive' very large aircraft and provide simulated feedback and sensor interpretation there to pilots. Same for trains, trams, vehicles in factories and warehouses... So why not cars? ;)
More information about the use-livecode