ISF Logo   IS Forum
Forum Index Register Members List Events Mark Forums Read Help

Go Back   International Skeptics Forum » General Topics » Science, Mathematics, Medicine, and Technology
 


Welcome to the International Skeptics Forum, where we discuss skepticism, critical thinking, the paranormal and science in a friendly but lively way. You are currently viewing the forum as a guest, which means you are missing out on discussing matters that are of interest to you. Please consider registering so you can gain full use of the forum features and interact with other Members. Registration is simple, fast and free! Click here to register today.
Reply
Old 26th March 2018, 04:31 AM   #281
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 74,711
Originally Posted by JoeMorgue View Post
My point, for all the difference it makes at this point, is that people are using "It needs to be safer!" when they mean "I'm not comfortable with it."
That might be your point, but since people who are also comfortable with it are saying the same thing, you can't claim to be able to distinguish between the two groups.

Quote:
Automated cars are already safer than human drivers. That's... one of those "fact" things.
The reality, however, is that humans judge safety differently depending on whether they are in control of the vehicle or not. That's... one of those "human" things.

Thing is, robots are better at a number of tasks but humans can still compensate for the odd lapse if they have the time.

As I said earlier, in this thread or another, sometimes when you're driving you can just "tell" that another driver is about to change lanes. I don't know what data the brain uses to make that determination, but since I am not aware of the process, I'd be hard-pressed to program a car to have the same "instict". I've lost count of the number of times I've said "I knew it" on the road when someone does a stupid maneuver.
__________________
Master of the Shining Darkness

Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 04:39 AM   #282
shuttlt
Philosopher
 
Join Date: Aug 2008
Posts: 5,631
Originally Posted by Belz... View Post
I think it's safe to say that if a human driver doesn't even _try_ to avoid the pedestrian, he doesn't deserve to be on the road. A robot driver that does the same needs to be improved upon.
We all have lapses of concentration while driving. If you restricted driving to people who were so careful this could not happen, the roads would be almost empty.

Sent from my Vodafone Smart ultra 6 using Tapatalk
shuttlt is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 04:45 AM   #283
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 74,711
Originally Posted by Klimax View Post
They are testing in (almost) best conditions. By definition they are far from ready. heavy rain/snow, mountain roads, roads in old inner cities of Europe,... There is far more fun ahead of them.
I've driven in conditions where the road was invisible to me because of blowing snow (do you not have a better term for that in English?), and where I had to use my knowledge of the area to get where I was going. Wasn't a problem. I doubt a computer could've done it without seeing the road at all.

Originally Posted by GlennB View Post
For my money at least attempting to stop in such a situation is so fundamental it gets sorted out in the lab and test track, not on public roads.
Thing for me is, when a human driver is not distracted somehow, they will react consistently to a situation. Though robots are, I'm sure, safer on the whole, they do have weird lapses sometimes where they just don't react to a situation at all. Sensors didn't work, or weren't interpreted properly, etc. That's not an issue with humans.

Originally Posted by jimbob View Post
2. LIDAR is Laser Infra-red based so visible colours are unimportant, just IR reflectivity.
So what happens if I'm wearing one of those volcano protective suits?
__________________
Master of the Shining Darkness

Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 04:48 AM   #284
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 74,711
Originally Posted by smartcooky View Post
If you think that pedestrians crossing the road, at night, pushing a bicycle with no reflectors, wearing dark clothing, on a four-lane open highway, not on a crosswalk, is not an edge case, then I would like you to tell me what is!
People running across the street at night while wearing black is common enough that I always look for it when driving past daylight. That's not an edge case.

Originally Posted by shuttlt View Post
We all have lapses of concentration while driving.
That's besides the point. I'm talking about a human paying attention to the road, and a robot shouldn't have these lapses.

Quote:
If you restricted driving to people who were so careful this could not happen, the roads would be almost empty.
And a lot safer for people like me, who do pay attention.
__________________
Master of the Shining Darkness

Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 05:35 AM   #285
jimbob
Uncritical "thinker"
 
jimbob's Avatar
 
Join Date: Jan 2007
Location: UK
Posts: 19,012
Originally Posted by Belz... View Post
I've driven in conditions where the road was invisible to me because of blowing snow (do you not have a better term for that in English?), and where I had to use my knowledge of the area to get where I was going. Wasn't a problem. I doubt a computer could've done it without seeing the road at all.



Thing for me is, when a human driver is not distracted somehow, they will react consistently to a situation. Though robots are, I'm sure, safer on the whole, they do have weird lapses sometimes where they just don't react to a situation at all. Sensors didn't work, or weren't interpreted properly, etc. That's not an issue with humans.



So what happens if I'm wearing one of those volcano protective suits?
You will stand out like nobody's business. They are highly reflective.
__________________
OECD healthcare spending
Expenditure on healthcare
http://www.oecd.org/els/health-systems/health-data.htm
link is 2015 data (2013 Data below):
UK 8.5% of GDP of which 83.3% is public expenditure - 7.1% of GDP is public spending
US 16.4% of GDP of which 48.2% is public expenditure - 7.9% of GDP is public spending
jimbob is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 05:48 AM   #286
Disbelief
Master Poster
 
Disbelief's Avatar
 
Join Date: Mar 2007
Posts: 2,599
Originally Posted by Klimax View Post
They are testing in (almost) best conditions. By definition they are far from ready. heavy rain/snow, mountain roads, roads in old inner cities of Europe,... There is far more fun ahead of them.
For this application, they are not far from ready. They are doing this testing in the locations they've chosen, specifically for the conditions, because they know the capabilities of the systems. They aren't just going to suddenly drop autonomous vehicles into the system that are perfect in every weather condition, so they will build information by testing where they are capable.
__________________
Zensmack (LastChild, Laughing Assassin, RazetheFlag, Wastrel, TruthbyDecree) - Working his way up the sock puppet chain, trying to overtake P'Doh. Or, are they the same?

Quote me where I said conspiracists use evidence. - mchapman
Disbelief is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 06:05 AM   #287
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 74,711
Originally Posted by jimbob View Post
You will stand out like nobody's business. They are highly reflective.
Good. Next purchase, then.
__________________
Master of the Shining Darkness

Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 06:28 AM   #288
rwguinn
Penultimate Amazing
 
rwguinn's Avatar
 
Join Date: Apr 2003
Location: 16 miles from 7 lakes
Posts: 11,096
Originally Posted by smartcooky View Post
If you think that pedestrians crossing the road, at night, pushing a bicycle with no reflectors, wearing dark clothing, on a four-lane open highway, not on a crosswalk, is not an edge case, then I would like you to tell me what is!

ETA: If people are looking for a reason why she was not detected, maybe its because she wasn't supposed to be... a bike presenting a side on view in the inside lane could have been a car... a normal occurrence; that lane is supposed to have cars in it. Of course, it still should have been detected if it was swerving into the lane, but that would be a lot harder to detect than a pedestrian walking in from the wrong side of the road.
A person pushing a bicycle, which was festooned with bags, suitcases, and/or all kinds of personal possessions, positioned perpendicular to the car's direction of travel could well present as a small car on the other lane.
It's velocity into the current driving lane would be pretty low, and could be interpreted as normal lane wandering.
The real problem is that it's relative speed wrt the car would be whatever the car is doing; i.e., it is not actually moving forward. This presents a problem of "if it's not moving, why is it wandering in its lane?", which apparently wasn't a scenario it was prepared to deal with.
It's a hypothesis...
__________________
"Political correctness is a doctrine,...,which holds forth the proposition that it is entirely possible to pick up a turd by the clean end."
"
I pointed out that his argument was wrong in every particular, but he rightfully took me to task for attacking only the weak points." Myriad http://forums.randi.org/showthread.php?postid=6853275#post6853275
rwguinn is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 06:53 AM   #289
Klimax
NWO Cyborg 5960x (subversion VPUNPCKHQDQ)
 
Klimax's Avatar
 
Join Date: Feb 2008
Location: Starship Wanderer - DS9
Posts: 12,527
Originally Posted by smartcooky View Post
If you think that pedestrians crossing the road, at night, pushing a bicycle with no reflectors, wearing dark clothing, on a four-lane open highway, not on a crosswalk, is not an edge case, then I would like you to tell me what is!
Sudden appearance of couch on road. Or similar highly unlikely event. (Say sudden landing of Statue of Liberty painted in black on highway) Still you must be able to avoid crashing into such thing. Slow pedestrian from distant median is trivial case to avoid in comparison.

Quote:
ETA: If people are looking for a reason why she was not detected, maybe its because she wasn't supposed to be... a bike presenting a side on view in the inside lane could have been a car... a normal occurrence; that lane is supposed to have cars in it. Of course, it still should have been detected if it was swerving into the lane, but that would be a lot harder to detect than a pedestrian walking in from the wrong side of the road.
Absolutely doesn't matter. Totally irrelevant in extreme. Something tries to cross road, you must be able to avoid it. Especially if "you" is computer with far better sensor suite. Or at least supposedly...

Originally Posted by Pixel42 View Post
That's a good point. I'd been assuming that the car didn't brake because it failed to detect the woman pushing her bike at all, which was puzzling. Detecting and misidentifying her as a car in the overtaking lane going in the same direction, and hence not a reason to take evasive action, does seem more likely.
In that case, AVs are even less ready then thought. Not only they cannot avoid some object that is moving across vehicles path, but they actively really on identification of said object to assess simple dangerous situation.

Originally Posted by shuttlt View Post
We don't know that it detected a substantial thing approaching the path of the car.

Sent from my Vodafone Smart ultra 6 using Tapatalk
If it was blind or not having enough time for reaction, then it is dangerous thing and needs to be taken off roads. (Drives dangerously fast and is blind)

Originally Posted by Disbelief View Post
For this application, they are not far from ready. They are doing this testing in the locations they've chosen, specifically for the conditions, because they know the capabilities of the systems. They aren't just going to suddenly drop autonomous vehicles into the system that are perfect in every weather condition, so they will build information by testing where they are capable.
It is highly unlikely that AVs will be ready (for whatever value of "ready") for decades. Most likely we will need true AI to actually have good AVs. And we are still quite far from that.
__________________
ModBorg

Engine: Ibalgin 400
Klimax is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 06:58 AM   #290
Klimax
NWO Cyborg 5960x (subversion VPUNPCKHQDQ)
 
Klimax's Avatar
 
Join Date: Feb 2008
Location: Starship Wanderer - DS9
Posts: 12,527
Originally Posted by rwguinn View Post
A person pushing a bicycle, which was festooned with bags, suitcases, and/or all kinds of personal possessions, positioned perpendicular to the car's direction of travel could well present as a small car on the other lane.
It's velocity into the current driving lane would be pretty low, and could be interpreted as normal lane wandering.
The real problem is that it's relative speed wrt the car would be whatever the car is doing; i.e., it is not actually moving forward. This presents a problem of "if it's not moving, why is it wandering in its lane?", which apparently wasn't a scenario it was prepared to deal with.
It's a hypothesis...
Ehm, movement was clearly from one side to another (she should have eben visible to LIDAR when she was still on median), even if at very slow speed. It still should have triggered braking because projection would show that said "car" is likely to cross into AVs lane. It doesn't matter if it is wrong for "car" to make such move. (For example there could have been sudden loss of control or something like that)
__________________
ModBorg

Engine: Ibalgin 400
Klimax is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 07:05 AM   #291
bruto
Penultimate Amazing
 
bruto's Avatar
 
Join Date: Jun 2005
Location: Way way north of Diddy Wah Diddy
Posts: 23,028
Originally Posted by shuttlt View Post
Everybody seems to be looking at this at quite a high level. Certainly any self drive car needs to be able to detect people in its path regardless of whether it is reasonable for them to be there. However, it can't directly detect people, so its not that easy. It has a bunch of inputs which may sometimes contradict each other and will contain varying degrees of noise. They can't solve the problem by just turning the sensitivity way up, or they would emergency break every time a gust of wind blew leaves into the road. It is clearly the case that even Uber detects objects in their path most of the time, or all their cars would be off the road due to accidents already. Hence, this is an edge case where the algorithm failed to appropriately handle some unusual set of inputs. The fact that the outcome was really bad does not make it not an edge case. I don't see how one could reasonably expect them to catch all the ways these systems could fail horribly on a test track.

Planes fail horribly sometimes despite huge amounts of testing, and the fact that not crashing into the ground at high speed or breaking up in the air being obvious things that we don't want to happen.

Sent from my Vodafone Smart ultra 6 using Tapatalk
That is an interesting point, but it also raises the question of the degree to which a case that might be fairly routine for a human driver is inherently an edge case for a self driving car, and whether there is any solution possible. And that, in turn, raises the question of whether the overall improvement in safety will be at the cost of local pockets of worsening.

I still think it's necessary to consider, not only the overall statistic of safety, but what form it takes. In the overall count of highway deaths, a sober robocar that does not fall asleep at the wheel and does not wander into the opposing lane, and reliably avoids a host of other stupid mistakes, will likely be statistically safer even if, owing to a limitation, not of implementation but of design, it picks off more pedestrians than ever.

And who knows, maybe that's a reasonable price to pay. In a sense, after all, it's a consequence of irresponsible behavior, and when you do it, you count unduly on the ability of others to compensate. So don't jaywalk because the robocars won't cut you any slack. It's an argument that can be made, but if so, it cannot be made without acknowledging, not only that it is the case but that it has to be.
__________________
I love this world, but not for its answers. (Mary Oliver)

Quand il dit "cuic" le moineau croit tout dire. (When he's tweeted the sparrow thinks he's said it all. (Jules Renard)
bruto is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 07:51 AM   #292
Disbelief
Master Poster
 
Disbelief's Avatar
 
Join Date: Mar 2007
Posts: 2,599
Originally Posted by Klimax View Post
It is highly unlikely that AVs will be ready (for whatever value of "ready") for decades. Most likely we will need true AI to actually have good AVs. And we are still quite far from that.
Won't even be close to that long.
__________________
Zensmack (LastChild, Laughing Assassin, RazetheFlag, Wastrel, TruthbyDecree) - Working his way up the sock puppet chain, trying to overtake P'Doh. Or, are they the same?

Quote me where I said conspiracists use evidence. - mchapman
Disbelief is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 08:03 AM   #293
shuttlt
Philosopher
 
Join Date: Aug 2008
Posts: 5,631
Originally Posted by bruto View Post
That is an interesting point, but it also raises the question of the degree to which a case that might be fairly routine for a human driver is inherently an edge case for a self driving car, and whether there is any solution possible. And that, in turn, raises the question of whether the overall improvement in safety will be at the cost of local pockets of worsening.

I still think it's necessary to consider, not only the overall statistic of safety, but what form it takes. In the overall count of highway deaths, a sober robocar that does not fall asleep at the wheel and does not wander into the opposing lane, and reliably avoids a host of other stupid mistakes, will likely be statistically safer even if, owing to a limitation, not of implementation but of design, it picks off more pedestrians than ever.

And who knows, maybe that's a reasonable price to pay. In a sense, after all, it's a consequence of irresponsible behavior, and when you do it, you count unduly on the ability of others to compensate. So don't jaywalk because the robocars won't cut you any slack. It's an argument that can be made, but if so, it cannot be made without acknowledging, not only that it is the case but that it has to be.
The dream of the future aspect of this with self drive cars passing one another impossibly close at intersections clearly assumes a controlled environment without people. How can you do that if there are humans cycling, human drivers or pedestrians wandering into traffic? The assumption in a lot of the thinking is that in the near future humans will be outlawed from the roads, or will move to self-drive in such overwhelming numbers that it won't matter. Anything short of that makes life hard for self drive cars. They have to navigate an uncontrolled environment that isn't designed around their needs.
shuttlt is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 08:28 AM   #294
jimbob
Uncritical "thinker"
 
jimbob's Avatar
 
Join Date: Jan 2007
Location: UK
Posts: 19,012
Originally Posted by Klimax View Post
Sudden appearance of couch on road. Or similar highly unlikely event. (Say sudden landing of Statue of Liberty painted in black on highway) Still you must be able to avoid crashing into such thing. Slow pedestrian from distant median is trivial case to avoid in comparison.



Absolutely doesn't matter. Totally irrelevant in extreme. Something tries to cross road, you must be able to avoid it. Especially if "you" is computer with far better sensor suite. Or at least supposedly...



In that case, AVs are even less ready then thought. Not only they cannot avoid some object that is moving across vehicles path, but they actively really on identification of said object to assess simple dangerous situation.



If it was blind or not having enough time for reaction, then it is dangerous thing and needs to be taken off roads. (Drives dangerously fast and is blind)



It is highly unlikely that AVs will be ready (for whatever value of "ready") for decades. Most likely we will need true AI to actually have good AVs. And we are still quite far from that.
In support of the first two paragraphs.

In some countries, large animals crossing the road are already significant causes of crashes - remember the Mercedes A class "moose test" fails.

I mentioned upthread the occasion where I was driving on the M40 in the dark at 70mph and saw a lorry tyre in one of the lanes. Unexpected obstacles do occur.
__________________
OECD healthcare spending
Expenditure on healthcare
http://www.oecd.org/els/health-systems/health-data.htm
link is 2015 data (2013 Data below):
UK 8.5% of GDP of which 83.3% is public expenditure - 7.1% of GDP is public spending
US 16.4% of GDP of which 48.2% is public expenditure - 7.9% of GDP is public spending
jimbob is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 08:36 AM   #295
jimbob
Uncritical "thinker"
 
jimbob's Avatar
 
Join Date: Jan 2007
Location: UK
Posts: 19,012
Originally Posted by Belz... View Post

As I said earlier, in this thread or another, sometimes when you're driving you can just "tell" that another driver is about to change lanes. I don't know what data the brain uses to make that determination, but since I am not aware of the process, I'd be hard-pressed to program a car to have the same "instict". I've lost count of the number of times I've said "I knew it" on the road when someone does a stupid maneuver.
Actually, machine learning algorithms are probably highly capable of learning such cues.

Unfortunately it's very difficult to assess what correlations have been made as a result of the learning.

How that would play with assigning liability when there is a crash could thus be... problematic. Did the software engineers verify the algorithm sufficiently? Who could say?
__________________
OECD healthcare spending
Expenditure on healthcare
http://www.oecd.org/els/health-systems/health-data.htm
link is 2015 data (2013 Data below):
UK 8.5% of GDP of which 83.3% is public expenditure - 7.1% of GDP is public spending
US 16.4% of GDP of which 48.2% is public expenditure - 7.9% of GDP is public spending
jimbob is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 09:13 AM   #296
Meadmaker
Penultimate Amazing
 
Meadmaker's Avatar
 
Join Date: Apr 2004
Posts: 18,298
Originally Posted by Klimax View Post
Ehm, movement was clearly from one side to another (she should have eben visible to LIDAR when she was still on median), even if at very slow speed. It still should have triggered braking because projection would show that said "car" is likely to cross into AVs lane. It doesn't matter if it is wrong for "car" to make such move. (For example there could have been sudden loss of control or something like that)
In the "don't believe it" videos that were posted after the crash, humans were visible on each side of the road, and clearly visible as humans. If that car didn't see the human and recognize it as a human with plenty of time to stop, something went wrong with the vision processing, possibly including poor choice of camera sensitivity for the actual lighting conditions.
Meadmaker is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 09:51 AM   #297
rwguinn
Penultimate Amazing
 
rwguinn's Avatar
 
Join Date: Apr 2003
Location: 16 miles from 7 lakes
Posts: 11,096
Originally Posted by Klimax View Post
Ehm, movement was clearly from one side to another (she should have eben visible to LIDAR when she was still on median), even if at very slow speed. It still should have triggered braking because projection would show that said "car" is likely to cross into AVs lane. It doesn't matter if it is wrong for "car" to make such move. (For example there could have been sudden loss of control or something like that)
I thought that was my point. The "Why " of recognition failure is actually a vital point in resolving the problem.
A "vehicle " with zero forward motion and very low lateral motion should trigger a braking and lane-change at the very least.
Hells, a STOPPED "vehicle" in an adjacent lane should trigger a lane-change action at the very least.
__________________
"Political correctness is a doctrine,...,which holds forth the proposition that it is entirely possible to pick up a turd by the clean end."
"
I pointed out that his argument was wrong in every particular, but he rightfully took me to task for attacking only the weak points." Myriad http://forums.randi.org/showthread.php?postid=6853275#post6853275
rwguinn is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 10:12 AM   #298
GlennB
Loggerheaded, earth-vexing fustilarian
 
GlennB's Avatar
 
Join Date: Sep 2006
Location: Arcadia, Greece
Posts: 23,603
Originally Posted by rwguinn View Post
I thought that was my point. The "Why " of recognition failure is actually a vital point in resolving the problem.
A "vehicle " with zero forward motion and very low lateral motion should trigger a braking and lane-change at the very least.
Hells, a STOPPED "vehicle" in an adjacent lane should trigger a lane-change action at the very least.
Quite. Whatever that thing is, it has no business being there and behaving that way. It's an "Oh ****!" moment, as it were.
__________________
"Even a broken clock is right twice a day. 9/11 truth is a clock with no hands." - Beachnut
GlennB is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 11:46 AM   #299
rwguinn
Penultimate Amazing
 
rwguinn's Avatar
 
Join Date: Apr 2003
Location: 16 miles from 7 lakes
Posts: 11,096
Originally Posted by GlennB View Post
Quite. Whatever that thing is, it has no business being there and behaving that way. It's an "Oh ****!" moment, as it were.
Actually, it was a "what the hell?" that got turned into an " Oh, ******" moment when it should have been a "I wonder what that was all about?"
__________________
"Political correctness is a doctrine,...,which holds forth the proposition that it is entirely possible to pick up a turd by the clean end."
"
I pointed out that his argument was wrong in every particular, but he rightfully took me to task for attacking only the weak points." Myriad http://forums.randi.org/showthread.php?postid=6853275#post6853275
rwguinn is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 12:02 PM   #300
bruto
Penultimate Amazing
 
bruto's Avatar
 
Join Date: Jun 2005
Location: Way way north of Diddy Wah Diddy
Posts: 23,028
Originally Posted by shuttlt View Post
The dream of the future aspect of this with self drive cars passing one another impossibly close at intersections clearly assumes a controlled environment without people. How can you do that if there are humans cycling, human drivers or pedestrians wandering into traffic? The assumption in a lot of the thinking is that in the near future humans will be outlawed from the roads, or will move to self-drive in such overwhelming numbers that it won't matter. Anything short of that makes life hard for self drive cars. They have to navigate an uncontrolled environment that isn't designed around their needs.
In the other robocar thread I argued to the point of figurative hoarseness about this issue but the general response was that technology would catch up and it would all be all right before you can say boo. But living in the country as I do, I foresee a long road ahead before a self driving car can do anywhere near as good a job figuring out what to stop for and what not to stop for.

As an example, having driven in the country for over 50 years, I have encountered many things on the road, and, owing sometimes to inevitability and sometimes to evaluative decisions, I have run over a few of them, including a few biggish animals that I decided were not worth the risk. I've run over a few woodchucks and porcupines, hit a couple of deer, and once a very fast dog on a crowded highway. One of the things I did not run over once was an escaped toddler. I slammed on the brakes, swerved, left the car idling in the middle of the road and snatched the baby back to the house where its hugely embarrassed parent had fallen asleep. It's a very rare instance, and one that is statistically insignificant to the vanishing point, and yet I surmise that it was pretty damned important at the local level.

What leads you to stop for something you see on or off the road? What are the cues and clues that make you run over a possum but not a cat? Stop for a smoldering school bus but not a log truck? Swerve for a bicycle even in traffic, but go ahead and hit that shopping cart? See the tire on the trailer truck ahead shredding before it sprays shrapnel into your windshield?

Chances are we can all, after a reasonable time driving, point to instances like this where we guessed right, foresaw something or noticed an anomaly, saw a pair of eyes in the woods or a shadow crossing a distant light source, or just knew that this is a curve where you'd better expect oncoming traffic, and so forth. And we can all be sure that from the overall point of statistical significance, compared to the many stupid mistakes that many drivers make, these things are statistically trivial. And yet, they seem oddly important at the individual level.

We live in a world whose outcomes are hugely impacted by many mistakes, some inevitable, but many stupid and careless, mitigated by relatively fewer instances where intelligence prevented or altered them. It's entirely reasonable to look for ways to prevent everyone including ourselves from having to pay the price for the sins and mistakes of others, but it may not be as reasonable to create a world where not only the mistakes but the mitigations are removed.
__________________
I love this world, but not for its answers. (Mary Oliver)

Quand il dit "cuic" le moineau croit tout dire. (When he's tweeted the sparrow thinks he's said it all. (Jules Renard)

Last edited by bruto; 26th March 2018 at 12:05 PM.
bruto is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 12:19 PM   #301
GlennB
Loggerheaded, earth-vexing fustilarian
 
GlennB's Avatar
 
Join Date: Sep 2006
Location: Arcadia, Greece
Posts: 23,603
Originally Posted by bruto View Post
In the other robocar thread I argued to the point of figurative hoarseness about this issue but the general response was that technology would catch up and it would all be all right before you can say boo. But living in the country as I do, I foresee a long road ahead before a self driving car can do anywhere near as good a job figuring out what to stop for and what not to stop for.

As an example, having driven in the country for over 50 years, I have encountered many things on the road, and, owing sometimes to inevitability and sometimes to evaluative decisions, I have run over a few of them, including a few biggish animals that I decided were not worth the risk...
Amen. You can tell without calculation whether that small animal is a child or not and make the appropriate judgement. Programming anything resembling 'judgement' into a machine is a real bitch.

For the foreseeable AV future, the software will have to default to 'ultimate safety mode, in case it's a human'. In the case of this accident it failed to do even that, assuming it even detected the danger.
__________________
"Even a broken clock is right twice a day. 9/11 truth is a clock with no hands." - Beachnut
GlennB is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 12:24 PM   #302
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 74,711
Right. Sometimes it's safer to run over a small animal than trying to avoid it. You don't have the same reaction to a toddler.
__________________
Master of the Shining Darkness

Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 01:00 PM   #303
JimOfAllTrades
Critical Thinker
 
Join Date: Aug 2011
Posts: 442
Originally Posted by rwguinn View Post
I thought that was my point. The "Why " of recognition failure is actually a vital point in resolving the problem.
A "vehicle " with zero forward motion and very low lateral motion should trigger a braking and lane-change at the very least.
Hells, a STOPPED "vehicle" in an adjacent lane should trigger a lane-change action at the very least.
The highlighted isn't always, or even most of the time, true.

For example, an ordinary city road with two lanes in each direction. At an intersection with a stop light there is an extra lane on the right for turning onto the crossroad. It will be very common for one or both straight through lanes to have stopped cars in them while the cars in right hand turn lane should continue past those cars up to the stopping point for the turn.

Or cars could be stopped between intersections because of construction, an accident, or just heavy traffic. But if your lane is clear you would continue past the stopped cars (albeit carefully).

However, if there is a vehicle in an adjacent lane with some side motion into your lane I would think that should be a lane-change or stop, regardless of forward motion.
JimOfAllTrades is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 01:42 PM   #304
shuttlt
Philosopher
 
Join Date: Aug 2008
Posts: 5,631
Originally Posted by bruto View Post
In the other robocar thread I argued to the point of figurative hoarseness about this issue but the general response was that technology would catch up and it would all be all right before you can say boo. But living in the country as I do, I foresee a long road ahead before a self driving car can do anywhere near as good a job figuring out what to stop for and what not to stop for.

As an example, having driven in the country for over 50 years, I have encountered many things on the road, and, owing sometimes to inevitability and sometimes to evaluative decisions, I have run over a few of them, including a few biggish animals that I decided were not worth the risk. I've run over a few woodchucks and porcupines, hit a couple of deer, and once a very fast dog on a crowded highway. One of the things I did not run over once was an escaped toddler. I slammed on the brakes, swerved, left the car idling in the middle of the road and snatched the baby back to the house where its hugely embarrassed parent had fallen asleep. It's a very rare instance, and one that is statistically insignificant to the vanishing point, and yet I surmise that it was pretty damned important at the local level.

What leads you to stop for something you see on or off the road? What are the cues and clues that make you run over a possum but not a cat? Stop for a smoldering school bus but not a log truck? Swerve for a bicycle even in traffic, but go ahead and hit that shopping cart? See the tire on the trailer truck ahead shredding before it sprays shrapnel into your windshield?

Chances are we can all, after a reasonable time driving, point to instances like this where we guessed right, foresaw something or noticed an anomaly, saw a pair of eyes in the woods or a shadow crossing a distant light source, or just knew that this is a curve where you'd better expect oncoming traffic, and so forth. And we can all be sure that from the overall point of statistical significance, compared to the many stupid mistakes that many drivers make, these things are statistically trivial. And yet, they seem oddly important at the individual level.

We live in a world whose outcomes are hugely impacted by many mistakes, some inevitable, but many stupid and careless, mitigated by relatively fewer instances where intelligence prevented or altered them. It's entirely reasonable to look for ways to prevent everyone including ourselves from having to pay the price for the sins and mistakes of others, but it may not be as reasonable to create a world where not only the mistakes but the mitigations are removed.
I was just idly wondering about software updates. Google and Microsoft every once in a while take out half a continent while updating a few routers. If self drive takes off, the first major mistake in an update to the self drive software is going to be interesting. Might give "race condition" a new meaning.

In any case, I wonder about the economics. If everybody quickly leaps on board with ordering a self drive car on their phone when they need one, and gives up on owning a car... OK. If not, do we know how much self drive cars will cost? Uber and haulage companies are clearly going to want this, but the car I'm currently in I thought twice about the cost of adaptive cruise control... If self drive costs more than that, I'm not sure that I would have gone for it.

Sent from my Vodafone Smart ultra 6 using Tapatalk
shuttlt is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 02:44 PM   #305
Metullus
Forum ¾-Wit Pro Tem
 
Metullus's Avatar
 
Join Date: Jan 2005
Posts: 5,038
I am curious, and I apologize if this has been addressed before, but would we expect the car to have identified the pedestrian in daylight? And if so, why?
__________________
I have met Tim at TAM. He is of sufficient height to piss on your leg. - Doubt 10/7/2005 - I'll miss Tim.

Aristotle taught that the brain exists merely to cool the blood and is not involved in the process of thinking. This is true only of certain persons. - Will Cuppy
Metullus is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 03:00 PM   #306
jimbob
Uncritical "thinker"
 
jimbob's Avatar
 
Join Date: Jan 2007
Location: UK
Posts: 19,012
Originally Posted by Metullus View Post
I am curious, and I apologize if this has been addressed before, but would we expect the car to have identified the pedestrian in daylight? And if so, why?
It would have increased the illumination and made the car visual systems more effective* but the LIDAR and radars should have been reliable enough anyway, otherwise the car hadno business being on the road at night.


* I have used monochrome machine vision systems in training production machines to identify locations on silicon wafers, and illumination helps there about as much as you might expect.
__________________
OECD healthcare spending
Expenditure on healthcare
http://www.oecd.org/els/health-systems/health-data.htm
link is 2015 data (2013 Data below):
UK 8.5% of GDP of which 83.3% is public expenditure - 7.1% of GDP is public spending
US 16.4% of GDP of which 48.2% is public expenditure - 7.9% of GDP is public spending
jimbob is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 04:27 PM   #307
Metullus
Forum ¾-Wit Pro Tem
 
Metullus's Avatar
 
Join Date: Jan 2005
Posts: 5,038
Originally Posted by jimbob View Post
It would have increased the illumination and made the car visual systems more effective* but the LIDAR and radars should have been reliable enough anyway, otherwise the car hadno business being on the road at night.


* I have used monochrome machine vision systems in training production machines to identify locations on silicon wafers, and illumination helps there about as much as you might expect.
Would daylight have prevented the accident (without human intervention)?
__________________
I have met Tim at TAM. He is of sufficient height to piss on your leg. - Doubt 10/7/2005 - I'll miss Tim.

Aristotle taught that the brain exists merely to cool the blood and is not involved in the process of thinking. This is true only of certain persons. - Will Cuppy
Metullus is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 06:39 PM   #308
bruto
Penultimate Amazing
 
bruto's Avatar
 
Join Date: Jun 2005
Location: Way way north of Diddy Wah Diddy
Posts: 23,028
Originally Posted by Metullus View Post
Would daylight have prevented the accident (without human intervention)?
If the accident happened in daylight, I don't think there would be any room for debate. The pedestrian was crossing an open highway, in steady motion, and wheeling a bicycle. Visibility was clearly unobscured. If the car could not detect that large an object, it's reasonable to assume it was far from adequate, and did not belong on the road at all. I don't think it's so important whether the car could identify her. There should have been plenty of visibility in plenty of time, and the object was far too big to run over safely. In those circumstances the car should do its thing whether the object in question is a person, a deer, or a runaway tire.
__________________
I love this world, but not for its answers. (Mary Oliver)

Quand il dit "cuic" le moineau croit tout dire. (When he's tweeted the sparrow thinks he's said it all. (Jules Renard)

Last edited by bruto; 26th March 2018 at 06:43 PM.
bruto is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 09:29 PM   #309
smartcooky
Philosopher
 
smartcooky's Avatar
 
Join Date: Oct 2012
Location: Nelson, New Zealand
Posts: 9,204
Originally Posted by Belz... View Post
As I said earlier, in this thread or another, sometimes when you're driving you can just "tell" that another driver is about to change lanes. I don't know what data the brain uses to make that determination, but since I am not aware of the process, I'd be hard-pressed to program a car to have the same "instict". I've lost count of the number of times I've said "I knew it" on the road when someone does a stupid maneuver.
Yep. A few weeks ago I was driving out to a town near where I live, on a 100km/h open road. As I approached an intersection, a car pulled up at a stop sign on the left (we drive on the left here and our cars are RHD). I realised that I hadn't seen his face in the driver's window, and just knew that he hadn't looked and was going to pull out in front of me; I started to brake just before he shot out onto the road. If I hadn't anticipated his actions, it would likely have been very nasty.
__________________
As democracy is perfected, the office represents, more and more closely, the inner soul of the people. We move toward a lofty ideal. On some great and glorious day the plain folks of the land will reach their heart's desire at last, and the White House will be adorned by a downright moron.
- Henry Louis Mencken - Baltimore Evening Sun, July 26, 1920

Last edited by smartcooky; 26th March 2018 at 09:40 PM.
smartcooky is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 09:48 PM   #310
Metullus
Forum ¾-Wit Pro Tem
 
Metullus's Avatar
 
Join Date: Jan 2005
Posts: 5,038
Originally Posted by bruto View Post
If the accident happened in daylight, I don't think there would be any room for debate. The pedestrian was crossing an open highway, in steady motion, and wheeling a bicycle. Visibility was clearly unobscured. If the car could not detect that large an object, it's reasonable to assume it was far from adequate, and did not belong on the road at all. I don't think it's so important whether the car could identify her. There should have been plenty of visibility in plenty of time, and the object was far too big to run over safely. In those circumstances the car should do its thing whether the object in question is a person, a deer, or a runaway tire.
If darkness was not a factor - if it did not impede the radar or LIDAR systems and had only a minimum impact on visual systems - then why are we debating whether or not a human driver would have reacted faster? Darkness would not be the issue. The same thing might well have happened in broad daylight.
__________________
I have met Tim at TAM. He is of sufficient height to piss on your leg. - Doubt 10/7/2005 - I'll miss Tim.

Aristotle taught that the brain exists merely to cool the blood and is not involved in the process of thinking. This is true only of certain persons. - Will Cuppy
Metullus is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 10:24 PM   #311
bruto
Penultimate Amazing
 
bruto's Avatar
 
Join Date: Jun 2005
Location: Way way north of Diddy Wah Diddy
Posts: 23,028
Originally Posted by Metullus View Post
If darkness was not a factor - if it did not impede the radar or LIDAR systems and had only a minimum impact on visual systems - then why are we debating whether or not a human driver would have reacted faster? Darkness would not be the issue. The same thing might well have happened in broad daylight.
I'm not sure what contribution visual sensors made to the whole. Since the accident n question did occur in darkness, it seems reasonable to compare the mechanical performance to a human one in the same conditions, and that is at least somewhat debatable. But indeed, if the system relied on non-visual cues, then it would have made the same error in daylight and the shortcoming would be more obvious.
__________________
I love this world, but not for its answers. (Mary Oliver)

Quand il dit "cuic" le moineau croit tout dire. (When he's tweeted the sparrow thinks he's said it all. (Jules Renard)
bruto is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 26th March 2018, 11:43 PM   #312
GlennB
Loggerheaded, earth-vexing fustilarian
 
GlennB's Avatar
 
Join Date: Sep 2006
Location: Arcadia, Greece
Posts: 23,603
Interesting to note that Tesla's 'Autopilot' hardware has a radar arc of ~45° forward-facing (only) and the rest of the long range kit is an array of cameras. The advertising blurb makes no mention of infra-red (and I could find no mention of IR elsewhere) though that doesn't preclude it being fitted. Tesla further claim that this is the final hardware setup required and full autonomy will be achieved through software upgrades.

Is a mere 45° of radar sufficient? Seems to me that if a Tesla and another vehicle are approaching a crossing on a collision course at 90° to each other, at the same speed and equidistant from the crossing then the Tesla will need 90° of radar to 'see' the other car. Other combinations of speed/distance could mean it needs more than 90°
__________________
"Even a broken clock is right twice a day. 9/11 truth is a clock with no hands." - Beachnut

Last edited by GlennB; 27th March 2018 at 01:07 AM.
GlennB is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 27th March 2018, 02:11 AM   #313
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 74,711
Originally Posted by smartcooky View Post
Yep. A few weeks ago I was driving out to a town near where I live, on a 100km/h open road. As I approached an intersection, a car pulled up at a stop sign on the left (we drive on the left here and our cars are RHD). I realised that I hadn't seen his face in the driver's window, and just knew that he hadn't looked and was going to pull out in front of me; I started to brake just before he shot out onto the road. If I hadn't anticipated his actions, it would likely have been very nasty.
Should've used the honk before he did, just to draw his attention to you.
__________________
Master of the Shining Darkness

Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 27th March 2018, 03:34 AM   #314
Klimax
NWO Cyborg 5960x (subversion VPUNPCKHQDQ)
 
Klimax's Avatar
 
Join Date: Feb 2008
Location: Starship Wanderer - DS9
Posts: 12,527
Originally Posted by bruto View Post
...
I still think it's necessary to consider, not only the overall statistic of safety, but what form it takes. In the overall count of highway deaths, a sober robocar that does not fall asleep at the wheel and does not wander into the opposing lane, and reliably avoids a host of other stupid mistakes, will likely be statistically safer even if, owing to a limitation, not of implementation but of design, it picks off more pedestrians than ever.

And who knows, maybe that's a reasonable price to pay. In a sense, after all, it's a consequence of irresponsible behavior, and when you do it, you count unduly on the ability of others to compensate. So don't jaywalk because the robocars won't cut you any slack. It's an argument that can be made, but if so, it cannot be made without acknowledging, not only that it is the case but that it has to be.
That's so American view. So wrong. Car above else. With such attitude, AVs won't ever be allowed in Europe.

Originally Posted by Disbelief View Post
Won't even be close to that long.
Such optimist.

Originally Posted by rwguinn View Post
I thought that was my point. The "Why " of recognition failure is actually a vital point in resolving the problem.
A "vehicle " with zero forward motion and very low lateral motion should trigger a braking and lane-change at the very least.
Hells, a STOPPED "vehicle" in an adjacent lane should trigger a lane-change action at the very least.
OK. Misunderstood your post.

Originally Posted by Belz... View Post
Right. Sometimes it's safer to run over a small animal than trying to avoid it. You don't have the same reaction to a toddler.
Once a small bird landed right in front of my car. I had all eternity to ponder what to do (IIRC speed was 60 km/h). yet in the end I ran it over, because I couldn't go around due to traffic and I couldn't safely abruptly stop because of cars behind me.

Originally Posted by GlennB View Post
Interesting to note that Tesla's 'Autopilot' hardware has a radar arc of ~45° forward-facing (only) and the rest of the long range kit is an array of cameras. The advertising blurb makes no mention of infra-red (and I could find no mention of IR elsewhere) though that doesn't preclude it being fitted. Tesla further claim that this is the final hardware setup required and full autonomy will be achieved through software upgrades.

Is a mere 45° of radar sufficient? Seems to me that if a Tesla and another vehicle are approaching a crossing on a collision course at 90° to each other, at the same speed and equidistant from the crossing then the Tesla will need 90° of radar to 'see' the other car. Other combinations of speed/distance could mean it needs more than 90°
For some intersections here you'd need about 180° to safely spot other traffic. (And often it has right of way...)
__________________
ModBorg

Engine: Ibalgin 400
Klimax is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 27th March 2018, 04:03 AM   #315
rwguinn
Penultimate Amazing
 
rwguinn's Avatar
 
Join Date: Apr 2003
Location: 16 miles from 7 lakes
Posts: 11,096
General rule at speed (50+ mph): if it's taller than the front bumper, avoid! Brake, veer, whatever.
Lower than that, brake and steer away if there's no other traffic that you might hit.
That said, I brake hard even for squirrels.
__________________
"Political correctness is a doctrine,...,which holds forth the proposition that it is entirely possible to pick up a turd by the clean end."
"
I pointed out that his argument was wrong in every particular, but he rightfully took me to task for attacking only the weak points." Myriad http://forums.randi.org/showthread.php?postid=6853275#post6853275
rwguinn is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 27th March 2018, 04:22 AM   #316
Disbelief
Master Poster
 
Disbelief's Avatar
 
Join Date: Mar 2007
Posts: 2,599
Originally Posted by GlennB View Post
Interesting to note that Tesla's 'Autopilot' hardware has a radar arc of ~45° forward-facing (only) and the rest of the long range kit is an array of cameras. The advertising blurb makes no mention of infra-red (and I could find no mention of IR elsewhere) though that doesn't preclude it being fitted. Tesla further claim that this is the final hardware setup required and full autonomy will be achieved through software upgrades.

Is a mere 45° of radar sufficient? Seems to me that if a Tesla and another vehicle are approaching a crossing on a collision course at 90° to each other, at the same speed and equidistant from the crossing then the Tesla will need 90° of radar to 'see' the other car. Other combinations of speed/distance could mean it needs more than 90°
Not sure anyone believes that Tesla is remotely capable of level 5 autonomy - except Tesla and their slappies. Everyone else has multiple systems to accommodate various weather conditions and to attempt to be better than a human.
__________________
Zensmack (LastChild, Laughing Assassin, RazetheFlag, Wastrel, TruthbyDecree) - Working his way up the sock puppet chain, trying to overtake P'Doh. Or, are they the same?

Quote me where I said conspiracists use evidence. - mchapman
Disbelief is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 27th March 2018, 04:40 AM   #317
The Great Zaganza
Philosopher
 
The Great Zaganza's Avatar
 
Join Date: Aug 2016
Posts: 5,331
Originally Posted by Disbelief View Post
Not sure anyone believes that Tesla is remotely capable of level 5 autonomy - except Tesla and their slappies. Everyone else has multiple systems to accommodate various weather conditions and to attempt to be better than a human.

Tesla is collecting more real-life data about (self-)driving than arguably anyone else working on the problem.
I would wager that whatever expertise it is missing it could just buy.
__________________
Opinion is divided on the subject. All the others say it is; I say it isn’t.
The Great Zaganza is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 27th March 2018, 04:46 AM   #318
GlennB
Loggerheaded, earth-vexing fustilarian
 
GlennB's Avatar
 
Join Date: Sep 2006
Location: Arcadia, Greece
Posts: 23,603
Originally Posted by Disbelief View Post
Not sure anyone believes that Tesla is remotely capable of level 5 autonomy - except Tesla and their slappies. Everyone else has multiple systems to accommodate various weather conditions and to attempt to be better than a human.
Originally Posted by The Great Zaganza View Post
Tesla is collecting more real-life data about (self-)driving than arguably anyone else working on the problem.
I would wager that whatever expertise it is missing it could just buy.
I don't doubt they're collecting the data and can buy sensational software. What troubles me is their claim that this hardware setup is all you'll ever need. Better processing and data then gets the car to level 5 autonomy, they claim.
__________________
"Even a broken clock is right twice a day. 9/11 truth is a clock with no hands." - Beachnut
GlennB is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 27th March 2018, 04:56 AM   #319
Disbelief
Master Poster
 
Disbelief's Avatar
 
Join Date: Mar 2007
Posts: 2,599
Originally Posted by GlennB View Post
I don't doubt they're collecting the data and can buy sensational software. What troubles me is their claim that this hardware setup is all you'll ever need. Better processing and data then gets the car to level 5 autonomy, they claim.
And integrating new hardware into their system I can't believe would be that simple. They have nearly zero mile of true autonomous driving.
__________________
Zensmack (LastChild, Laughing Assassin, RazetheFlag, Wastrel, TruthbyDecree) - Working his way up the sock puppet chain, trying to overtake P'Doh. Or, are they the same?

Quote me where I said conspiracists use evidence. - mchapman
Disbelief is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 27th March 2018, 05:03 AM   #320
KAJ
New Blood
 
Join Date: Dec 2017
Posts: 6
Originally Posted by GlennB View Post
Interesting to note that Tesla's 'Autopilot' hardware has a radar arc of ~45° forward-facing (only) and the rest of the long range kit is an array of cameras. The advertising blurb makes no mention of infra-red (and I could find no mention of IR elsewhere) though that doesn't preclude it being fitted. Tesla further claim that this is the final hardware setup required and full autonomy will be achieved through software upgrades.

Is a mere 45° of radar sufficient? Seems to me that if a Tesla and another vehicle are approaching a crossing on a collision course at 90° to each other, at the same speed and equidistant from the crossing then the Tesla will need 90° of radar to 'see' the other car. Other combinations of speed/distance could mean it needs more than 90°
Draw the diagrams. It depends (inter alia) on distances from the junction. For right angle junctions you only need 90° when at the junction.
KAJ is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Reply

International Skeptics Forum » General Topics » Science, Mathematics, Medicine, and Technology

Bookmarks

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


All times are GMT -7. The time now is 12:52 PM.
Powered by vBulletin. Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.

This forum began as part of the James Randi Education Foundation (JREF). However, the forum now exists as
an independent entity with no affiliation with or endorsement by the JREF, including the section in reference to "JREF" topics.

Disclaimer: Messages posted in the Forum are solely the opinion of their authors.