In short, how does a computer logic out a solution when the input (Dog, cat, zombie pedestrian) uses no logic?
|
|
|
Odd thoughts on driverless cars
- Thread starter nobbie48
- Start date
In short, how does a computer logic out a solution when the input (Dog, cat, zombie pedestrian) uses no logic?
That's what the person from Mercedes was referring to in the article I linked to. The software will default to the solution that the car has the most control over the outcome (saving the passengers) then consider the other, possibly erratic factors in the equation.
What I read into it was someone at Mercedes "misspoke", and someone higher up the chain "clarified" their position. It's the Germans. They'll do what they intended to regardless, as they always have. eg: look at how long it took for the NA subsidiaries of VAG get the parent company to put cup holders in their cars. For years the company line was You shouldn't be drinking anything while driving, therefore cars don't need cup holders.
Last edited:
daught
Well-known member
People don't maintain their ebrake. It's very unlikely for both brakes to fail at the same time. Even if both fail electric/hybrid cars can brake using the engine. Anyways I have more faith in self driving cars behind me than most drivers.
Considering even the primary braking system is split into two (front and rear braking systems are intentionally separate so that the failure of one doesn't effect the other) you are now theoretically talking about the complete failure of 3 fully independent braking systems.
And commercial vehicles used for passengers are subject to annual safety inspections.
So yeah, not gonna happen.
Stop trying to bring this back to only reasonable things happening! We're trying to hash out our version of the three laws of robotics here, AND solve the ethics of the trolley problem. Oh, and real world, you can't stop a AMC Hornet, by shifting into park, but reverse does really well, especially if you remember to put your foot down before the car starts backing up. Eventually, some electric car will lose all its brakes. It's the law of averages.
JD_PICKLE
Well-known member
For navigation, do these driver-less cars use GPS? The reason I ask, I do know it's old, but my 5 year old Tom Tom GPS, while traveling Hwy #11 between Hearst and Longlac, very old hwy, been there a long time, the GPS said I was 300 meters off the road. Will a driver less car drive me through the forest rather than the road??
I know they have a lot to work out yet, but I hope not in my life time.
On a side note...
http://www.dailymail.co.uk/news/art...ot-car-strikes-police-motorcycle-Arizona.html
and....
https://newatlas.com/tesla-autopilot-fema/46045/
I know they have a lot to work out yet, but I hope not in my life time.
On a side note...
http://www.dailymail.co.uk/news/art...ot-car-strikes-police-motorcycle-Arizona.html
and....
https://newatlas.com/tesla-autopilot-fema/46045/
Most likely during our lifetime. Just look at how we went from 14k baud modems to LTE on our PHONES!
Advancements are happening fast. And in terms of GPS, definitely will be used for general navigation, but the sensors on the car are what are key here and what keeps you on the road.
I can imagine easily the car asking you if you wanna go "offroad" lol
You'd be hard pressed to find any stories out there of complete and total automobile brake failure. You are talking a literally infinitesimal risk scenario that literally has probably not happened to any vehicle out there since split braking systems (designed to avoid complete and total brake failure) became standard in the 1970's...
Yes, some people will "claim" their brakes failed when in reality they may have just lost their brake booster - the most common brake related issue in reality. There's actually a lawsuit on this exact issue against GM right now, but the plaintiffs claim "brake failure" but it's clear there brakes continue to work just fine with increased pedal pressure. In this scenario you still have perfectly functional brakes, you just need to push the pedal a lot harder to get them to work. Most people don't understand that and would just say they "failed", which wasn't the case.
A computer is far smarter than a human driver. If a booster fails, first off the computer will probably know about it and compensate perfectly find with increased pedal/system pressure seamlessly and automatically. If by some stretch then a brake line to the front brakes blows, a computer would then compensate again automatically by simply continuing to stop using the still perfectly functional rear brakes. Vice versa if the rear brakes fail, it would stop using the front brakes. If, by some incredible stretch of the imagination BOTH front and rear brake lines blew, then the computer would apply the physical E-Brake and stop that way instead.
ALL of the above scenarios are also perfectly viable options for a human driver, but in this era when people don't even understand how to turn on their own headlights if the automatic feature stops working, or drive around oblivious in hot sunny weather with their rear wiper flapping away (but for 2 examples), well, a lot of people would rush to scream "brake failure". Not the case, and a driverless car will be far smarter than a human in this situation.
The situation you're trying to present as an in inevitability is really anything but. Maybe in the pre 1960 era full vehicle braking failures were a thing, but they simply are not anymore inside the last 40-50 years. It simply does...not...happen. Especially on a commercial vehicle subject to inspection every 12 months.
Not trying to be a dick here, just stating facts.
It's all fun and games until the pedal goes right to the floor, and the car doesn't slow. The issue was a cracked master cylinder, and rusted emergency brake cable which I managed to snap. It's possible that they've changed things for the better, but there are so many vehicles on the road, that even a 0.001% or less chance event will happen sooner or later.
That's not very nice at all. Fortunate then, that I have no enemies.
We've had audible disk brake warnings for years and I'm sure the IT types can come up with a system for the driverless vehicles that gives you a 500 km to shutdown warning. On start up a high pressure test of hydraulics would be a go / no go for travel if a line ruptured etc. Same for the E-brake.
Because we don't have control over the vehicle we demand it to compensate for situations most drivers couldn't handle.
I read a short story about a high tech company secretly creating an android VIP to address the shareholders meeting, figuring that its massive and well programmed system could answer questions without making human judgement errors or tongue slips. At the meeting it eloquently recited sales figures and projections, profit analysis, liability issues and more.
All went well until a little old lady asked what kind of sandwiches were going to be served at the company picnic. The androids eyes popped out and it did a complete meltdown. What will be the Achilles heel of the DLV?
Some years back Bosch was having problems with their ABS and fuel injection systems failing when the vehicles encountered particular electrical fields, eddy currents and peculiar frequencies from radio towers.
When my daughter was learning to drive I mentioned total brake failure advising her to aim for something cheap and soft.
Bwahahaaahahahaha
https://www.theweathernetwork.com/news/articles/driverless-shuttle-bus-crash-las-vegas/89063/
Yeah, it wasn't the driverless bus at fault (the truck backed into it) but the inability of the driverless bus to think outside the box in which it was programmed incapacitated it. "Hmmm. This truck is maneuvering. I don't think the driver can see me. Maybe I should back up a little bit to give it some room. Or maybe I should toot the horn to alert the driver of my presence." Couldn't do that. Exceeds programmed capabilities.
https://www.theweathernetwork.com/news/articles/driverless-shuttle-bus-crash-las-vegas/89063/
Yeah, it wasn't the driverless bus at fault (the truck backed into it) but the inability of the driverless bus to think outside the box in which it was programmed incapacitated it. "Hmmm. This truck is maneuvering. I don't think the driver can see me. Maybe I should back up a little bit to give it some room. Or maybe I should toot the horn to alert the driver of my presence." Couldn't do that. Exceeds programmed capabilities.
People in driverless vehicles are going to get hurt, if their vehicles can't follow the rule of the most axles, in the short term.
Yep, classic case of religiously following the law not being a guarantee of safety. Though it is a guarantee of not being responsible, which may matter more to these corporations than safety.
This is all assuming a human driver would have anticipated the truck's need for space and avoided putting themselves in a position where they couldn't be seen from the cab. Reasonable assumptuI think.
Bet it won't on future on future software revisions.
That's the thing - we are in the infancy stage of driverless cars.
When steam engines first came out they were not problem free either - boilers exploded, people died, etc etc - but things got better and steam power became the defacto standard for decades. The first airplanes were death traps. The first generation of jet engines were known for blowing up. Early internal combustion engines were hardly reliable either. ABS was a nightmare when it was first introduced in the 60's and 70's, but now it's standard on every car out there.
My point is... all through all of these stepping stones I'm sure there were detractors rushing to point and laugh the same as how people are pointing and laughing about driverless cars now. That doesn't mean we should simply flush the technology down the toilet and forget about it, it means we just need to realize it's still a maturing technology and be a little bit patient.
O.k. Truck starts backing up. Automatic bus comes screeching to a halt, but too close. Bus is now programmed to back up, and needs to back up, because the truck is going to crush the passengers inside. Pedestrian steps out behind bus to jaywalk. Does the bus take him out? let the people inside die? Beep horn as an attempt to get attention? or something else. Lots of programming options.
If the bus backed up it would be at fault. If it got crushed while stationary it wouldn't be at fault. I think that's your answer.
I noticed nobody has commented on this.
This is actually the biggest legitimate issue with self driving cars.
I wrote scripts when I was much younger designed to crash systems. One removed everything from system32, disabled safe mode, put itself as a launch application, and eventually I figured out how to make it windows service. While I never let it loose, it wouldn't take a genius, back then, to attach the payload to a picture using the old .jpeg exploit.
You can bet someone is going to try and **** with self driving cars. Question is are they going to make them drift swastika patterns (somebody recently caused printers with exploits to print swastikas everywhere....this also happened a few years ago), or smash into pedestrians.
|
|
|
|
|