1 ... 8 9 10
SVreX (Forum Supporter)
SVreX (Forum Supporter) MegaDork
11/18/21 10:17 a.m.

Physics may not change, but real-world application of physics will always change as long as it is being applied in a natural environment. 

grpb
grpb Reader
11/18/21 11:27 a.m.

In reply to SVreX (Forum Supporter) :

I thought the human vs chess thing was settled long ago, it's just a question of how much the computer is handicapped when playing a human.

I think the basic question with computers is the same as with humans; What does 'learning' actually mean?  Humans lose at chess to computers because the 'rules' are rigid and unchanging, therefore the 'learning' that computers do is within a very narrow and specific realm.  To a computer, nothing outside of the chess board matters.  But, if the human cannot beat the computer within the confines of the chessboard, then humans will think of other ways, I dunno, some kind of denial of service attack at the access point of the computer playing, paying a friend to sneak in and tripping all the breakers in the building housing the computer, etc.  I would assume if somehow the computer cannot complete a move, that is a loss.  As a human, I would take that as a win, and nothing in the rules of chess specifies anything about circiuit breakers.  Is it in the 'spirit' of the game?  Maybe not, but we are goal driven creatures, not process driven.

Likewise with racing, a computer would do calculations based on what it knows the vehicle can do within the confines of the track.  Great human racers make calculations based on what is not too impossible, and that is the beauty of watching racing. Computers racing would end up like trains racing on the same track, and obviously the template to beat them is shown below.  Ciao Vale!!!!

https://youtu.be/sfPM77TsGaA?t=37

SVreX (Forum Supporter)
SVreX (Forum Supporter) MegaDork
11/18/21 11:45 a.m.

In reply to grpb :

The "rules" are rigid. Correct. 
 

But the real world of racing is not. No matter how many sensors are put on a car, it's still a changing environment. Air temperature, tire temperature, degradation of tire compounds, degradation of computer processing ability, different line, competitor's strategies, etc, etc.

Yes, I understand sensors can measure and calculate all those things. It's still a natural environment that changes, and each car will have unique programming parameters. 
 

If cars are gonna run at the absolute limit of physics, it could be the difference between one wire heating a little more than another. 
 

Racing is gonna be around for a long while. But it's gonna take different forms. 

SVreX (Forum Supporter)
SVreX (Forum Supporter) MegaDork
11/18/21 11:49 a.m.

Consider the battle bot competitions...

Machines. Each one designed to destroy the others. But there has never been one that was undefeatable. Every time someone invents a new method to defend or attack, someone comes up with a new idea. 
 

Same will happen with autonomous race cars for a very long time. 

frenchyd
frenchyd UltimaDork
11/18/21 12:39 p.m.

In reply to SVreX (Forum Supporter) :

Approaching a moving, rolling , sliding 400'x 100' landing area at 125 knots  at night  in a storm is the most exciting thing I've done in my life.   Miss and the life of the crew and myself was over,  the fuel left gives me one attempt.  
The LSO is giving me a wave off because it looks to him  like I'm not lined up at all and way too high.   What he doesn't see is that  big rogue wave over his shoulder that will push the stern of the carrier over and up about the time of touch down. 
     I touched down on the +2 wire centerline yet he gave me a 1.25 ( out of 4.0) for the landing. 
 That's thrilling. Satisfying to the soul.  
     Racing is almost like that.  I'm sure most people would avoid such risks.  But there is a significant portion of the population that crave  such sensation.
     No simulation, no war game, grand theft Auto has any way of providing that.  
   Auto pilot is now developed to the point where take off and landing is done by auto pilot. In reality  combat could now be handled by some tech pushing buttons   below deck. ( kinda the ultimate video game ) 

fidelity101 (Forum Supporter)
fidelity101 (Forum Supporter) UltraDork
11/18/21 2:52 p.m.

stage coach robberies will return

SVreX (Forum Supporter)
SVreX (Forum Supporter) MegaDork
11/18/21 4:44 p.m.

Ok. I give up. 
 

I'm trying to offer the best positive spins that I can. Don't think I can overcome the naysayers. 
 

So let me just say this... autonomous cars ARE coming. I WILL be racing them. I WILL have fun. 
 

Even if some of you are longing for your horse-drawn carriages. 

AnthonyGS (Forum Supporter)
AnthonyGS (Forum Supporter) GRM+ Memberand SuperDork
11/20/21 10:41 a.m.

It will happen.  The most dangerous part of auto travel is the inattentive or impaired person behind the wheel.  Even if that isn't you, it can still harm you.  And honestly, some people are convinced they can legislate everyone into a better life according to their vision whether you want it or not.

Self driving cars will become the norm; it's only a matter of time.  

frenchyd
frenchyd UltimaDork
11/20/21 10:37 p.m.
AnthonyGS (Forum Supporter) said:

It will happen.  The most dangerous part of auto travel is the inattentive or impaired person behind the wheel.  Even if that isn't you, it can still harm you.  And honestly, some people are convinced they can legislate everyone into a better life according to their vision whether you want it or not.

Self driving cars will become the norm; it's only a matter of time.  

Auto pilots have made air travel the safest way to travel.  Firm rules, rigidly enforced, with serious consequences for failure to follow those rules. 
   That's why Germany gets away with the autobahn.  Why Americans with our " rugged individualism" have trouble complying.  
     Road congestion, demands strict adherence. Safety systems can solve some issues with air bags, seat belts, crush zones etc. etc. but the more    "freedoms" on public roads the greater the fatalities. 

codrus (Forum Supporter)
codrus (Forum Supporter) GRM+ Memberand PowerDork
11/20/21 11:07 p.m.
frenchyd said:

Auto pilots have made air travel the safest way to travel. 

Auto pilots are not "self driving cars" or even "self flying planes".  Autopilots are basically fancy cruise control.

 

 

frenchyd
frenchyd UltimaDork
11/20/21 11:22 p.m.
codrus (Forum Supporter) said:
frenchyd said:

Auto pilots have made air travel the safest way to travel. 

Auto pilots are not "self driving cars" or even "self flying planes".  Autopilots are basically fancy cruise control.

 

 

Not the new stuff in Military  or commercial.  They have the ability to take off and land ( on carriers at sea)    Do in flight course corrections etc.  ( As a pilot who flew off and landed on Essex class carriers during Vietnam  I envy that ability )  

  I don't know how advanced auto pilot systems are on private planes but I can't imagine they are seriously behind,   at least in the private jets.  

codrus (Forum Supporter)
codrus (Forum Supporter) GRM+ Memberand PowerDork
11/21/21 12:04 a.m.
frenchyd said:

Not the new stuff in Military  or commercial.  They have the ability to take off and land ( on carriers at sea)    Do in flight course corrections etc.  ( As a pilot who flew off and landed on Essex class carriers during Vietnam  I envy that ability )  

The Navy has demonstrated an autonomous carrier landing ability in an X plane, but I don't get the impression it's a production feature in regular carrier aircraft (yet).

Some commercial jets have an autoland facility.  I am not a pilot, but from what I've read it requires specific instrument landing system equipment and will do a decent job of landing the plane in many circumstances (apparently humans are still smoother, but that's likely just a few software revisions to fix).  That's fine, but but when I say it's a "fancy cruise control" I mean that it does what you tell it.  It answers "how" to land a plane, but not "where", "when", or "why".  

The mechanics of landing a plane are the kind of relatively simple, well-defined problem that computers are good at.  Follow a radio beacon that gives the proper direction, maintain a particular descent rate, adjust the flight controls to keep the plane at the right speed and attitude, flare it at a certain height above the ground, etc etc.

What computers are NOT good at is open ended, abstract reasoning.  The left engine is running hot -- why is it doing that?  Does that mean we need to divert?  If so, where do we divert to?  Normally we'd go to airport X, but it's raining there, should we go to Y instead?  Oh look, there's a flock of birds at the end of the runway, do we need to go around to avoid them?  OMG that idiot in the 747 didn't hear the tower's instructions and just pulled onto the runway I'm trying to land on!

In some ways a self-driving car is a simpler problem than a self-flying airplane because it's not going to fall out of the sky.  In other ways it's actually more complicated because it's a much more dynamic environment.  There aren't many pedestrians jaywalking across a cloud, for example. :)  Also, the average driver is not paid or trained as well as the average pilot, and when you give him a "halfway" solution that requires him to take over when something weird happens, odds are that he's not paying attention and will miss it.

A truly autonomous car requires a whole different level of abstract reasoning to these kinds of "fancy cruise controls", and it's far from obvious how you make that happen.  You can't take the kinds of logic that you use to do an aircraft auto landing and just do more of it, there are far too many cases to enumerate them all.  IMHO, autonomously operating a car in the real world requires that the computer driving it be able to "think", and quite frankly we don't know how to make a computer do that.

 

Caperix
Caperix New Reader
11/21/21 6:21 a.m.

I think to have true self driving cars will require a set standard to allow car to car & car to road communication.  Even then if there are cars on the road that can do unexpected things there will be issues.

While the idea of sleeping through a long boaring car trip or watching Netflix on the rush hour drive home sounds good, I would still want to have a fun car to drive on the weekends.  If that car is under my control then it could cause an issue for self driving cars.

Right now self driving systems are fairly simple, lane keep to keep the car between the lines & active cruise to keep distance to the car in front of you. There is not any regulation to how it is done, some systems just use a camera some use camera & LIDAR.  True self driving has only been done under very controlled conditions, towns that have very little rain & have very little pedestrian traffic.

AnthonyGS (Forum Supporter)
AnthonyGS (Forum Supporter) GRM+ Memberand SuperDork
11/21/21 9:27 a.m.

In reply to Caperix :

We already have plenty of people watching Netflix on the commute home.  That's why self driving cars will happen.

frenchyd
frenchyd UltimaDork
11/21/21 11:53 a.m.

In reply to Caperix :

One recently watched video talked about carrier landing an F35 in auto.  
    Details about exactly how and what weren't present. But I do know carriers regularly go into endcom  where nothing is radiated from the ship.  Thus it has to be the plane landing itself passively without input from the ship. 

Erich
Erich UberDork
11/21/21 1:03 p.m.
Caperix said:

  True self driving has only been done under very controlled conditions, towns that have very little rain & have very little pedestrian traffic.

That's not exactly true. I live in a hotbed for testing autonomous cars - Ann Arbor, Michigan. We have quite a few companies prowling our streets with autonomous vehicles, a shuttle on the campus of University of Michigan, and a pilot program downtown with autonomous taxis.

The areas they've begun testing in have pretty heavy pedestrian traffic, are not at all controlled conditions, and from where I sit, it sure feels like we get our share of rain and snow.

I'll be anxious to see how they do. I've been a naysayer for a while on autonomous tech but I have a few friends who work in these companies and they're far more bullish on the prospects of getting to level 4 autonomy with extremely rare interventions, if any. They are much less bullish on Tesla's ability to get there. 

frenchyd
frenchyd UltimaDork
11/21/21 2:03 p.m.
Caperix said:

I think to have true self driving cars will require a set standard to allow car to car & car to road communication.  Even then if there are cars on the road that can do unexpected things there will be issues.

While the idea of sleeping through a long boaring car trip or watching Netflix on the rush hour drive home sounds good, I would still want to have a fun car to drive on the weekends.  If that car is under my control then it could cause an issue for self driving cars.

Right now self driving systems are fairly simple, lane keep to keep the car between the lines & active cruise to keep distance to the car in front of you. There is not any regulation to how it is done, some systems just use a camera some use camera & LIDAR.  True self driving has only been done under very controlled conditions, towns that have very little rain & have very little pedestrian traffic.

Someone will be the winner and someone the loser.  It's a political design now and money absolutely will muddy the waters. 
  Since the average politician barely can use a cell phone only if his staff set it up for him.  They will react how they always do in these matters, what's in it for me.  
    Thus there is no assurance we will get good or even workable regulations.  

1 ... 8 9 10
Our Preferred Partners
8bsJmLA7zT0SzHNnuQchgiZA3xO6pBVXYtmGle0xkl8CTeT56qsxvAQ9ZBHXZNDz