"What's your proposed model for testing an autonomous car driving amidst normal traffic conditions that does not include actually having it drive among normal road traffic?"
The obvious answer is to install the hardware and software on vehicles that are driven by humans performing normal driving and not to enable the autonomous control of the vehicle until millions of miles are accumulated with no false negatives (failure to detect a problem) and minimal false positives. In practice, that's not going to work perfectly. But it's clearly not what Uber is doing.
"Companies like Google and Facebook make it much easier to correlate all the data."
The irony being that almost all that data is actually quite useless. Sooner or later -- maybe next week, maybe a decade from now, folks will figure that out and there will be a massive market reevaluation.
So, We will end up with a block market selling older permanently lockable phones to those who do NOT wish to share their secrets with the FBI, KGB. NSA, local law enforcement and the weird kid down the street? Here's your chance to make a killing folks. Get in early. Business Plan? When a decsion is needed, just ask yourself -- "What would Uber Do?"
Naw, probably the police just assessed the situation as they would for a human driver. Nighttime. Pedestrian in a place a pedestrian shouldn't have been. An unimpaired human driver might well have been unable to avoid an accident and very likely wouldn't have been held responsible. BUT. The damn car seemingly SHOULD have done better than it did. Disturbingly it never seemed to try to avoid the accident even though it might have been too late to do so by the time the pedestrian was recognized. THAT seems very bad.
Conclusion: Probably not really ready for prime time.
The vehicle has LIDAR, right? Pedestrian dressed in black combined with night might well be a problem for the human driver but it shouldn't be a problem for the hardware? If it is a problem for the vehicle, The same situation is likely going to be a problem in broad daylight with the pedestrian dressed head to toe in dayglow orange, No?
FWIW, here's a link to the story in Plattsburgh's local newspaper which may provide a clearer explanation of the problem. http://www.pressrepublican.com... And yes, late December and early January did feature some unusually low temperatures -- around -17F(-27C) several nights presumably causing folks to run more electric heaters than usual more of the time.
I'm not a big fan of electric vehicles, And considering the price, I don't think anyone who much cares about cost per mile is likely to buy a Tesla. But we really shouldn't compare highway mileage for an ICE vehicle to EVs which I would assume will be used mostly for city driving and commuting applications where non-hybrid-ICE mileage tends to suck.
"The train is estimated to take 3 hours, compared to 5.5 hours by car."
Ten or Fifteen years ago, Mythbusters staged a race between a Bay Area starting point and an LA area destination with a pair driving a car and an individual taking a plane (nominally about an hour) then renting a car in LA. The racers arrived within a few minutes of one another. Their conclusion. Depending on luck, both take about the same time. My guess is that HSR would likely be about the same (fewer delays than air travel, but slower and won't sit at the airport for hours waiting for the fog to lift at LAX.
An even better example might be the Erie Canal which was built by New York State in the 1820s. NY built it after private capital refused to take the chance. The Erie Canal opened up settlement of mid-continent North America a couple of decades before the railroads got to be competetive. It cut travel time from the coast to the Great Lakes from a month to nine days. It turned places like Syracuse, Rochester and Buffalo into boom towns It cemented New York City's position as the nation's leading port. And it actually paid for itself.
That said, To me California' HSR project iooks like an (expensive) solution in search of a problem. I lived in California much of my life and I've also travelled a bit and lived overseas. My guess is that California probably needs a transportation system based on personal transportation, not -- with a few exceptions -- public transit. But maybe "cars" don't need need to be so big, intrusive, and generally obnoxious -- at least not in the Bay Area, LA Basin and San Diego. In the hinterland -- say Mono County -- I can't imagine what you'd replace them with.
Except that for political reasons, the newer, faster replacement is going to run through the distinctly not so scenic central valley rather than along the coast. And it's presumably going to cost a lot more than the Coast Daylight if they hope to ever recover construction costs. And there's a distinct -- once_you_get_to_X,_you_need_a_car_to_get_to_where_you_really_want_to_be problem in California's sprawling urban areas
I'm not against trains. I even ride them sometimes. But I think perhaps California needs clearer vision rather than a probably doomed attempt to imitate Europe, East Asia and the US East Coast corridor.
One thing that the slashdot description doesn't mention. Just having two transmitters in a room or building does not produce intermodulation products. You need some sort of non-linear "device" as well to combine the signals and emit the IM products -- in radio terms, a mixer. (That's not the same thing as an audio "mixer" BTW). I'm having a little trouble imagining what could act as a mixer and emit significant 7kHz audio signals. And wouldn't folks be able to find the mixer if not the transmitters just using their ears? 7KHz is well within the audio range for all but a few of us elderly folks.
In reality, the Cubans suggested that the US State Department to bring the FBI in to see if they could get to the bottom of the problem. (No, I don't know why the Cubans might think the FBI might know anything about sonic weapons).
Do the Cubans have the US offices and residences bugged? Of course they have the US offices and residences bugged. Why would they not? You can probably hear the plumbing talking to the air conditioners and every other electronic device if you listen real hard at night. Did they unleash some weird sonic weapon of the gringos? Even if they had such a device why would they use it on diplomats instead of on folks who are actually a threat?
It'd be more fun to make it look like an archway going to an adjacent room or a garden and watch people walk into it.
====
But seriously, I imagine that people with more money than sense -- no shortage of THOSE -- will find uses for these. Put up a famous picture -- The Nightwatch or September Morn. And change it occasionally.
Heck, there might even be some useful things these things can do.
Just wait until Ford or Toyota or VW inadvertently lets a certificate expire and a few hundred million vehicles glide silently to a stop... wherever they may happen to be,
"I shouldn't need to eat up bandwidth and route traffic between Vancouver and Cupertino simply to copy data between two devices 10 inches apart."
I shouldn't be so quick to broadcast that sort of obvious nonsense to the world citizen. Just keep in mind that without profits there is no progress, and that progress is the only reason for your continued existence. Next, you'll probably be asking questions like where it is, exactly, that you are progressing toward.
The todo.text format https://github.com/todotxt/tod... doesn't seem to have a simple date field. The closest I see is a FINISH START pair, each specified as a 10 character YYYY-MM-DD string. That's a lot of typing for a number that can be specified as YYMMDD. It's been a long time, but I'm pretty sure I looked at that and decided that the format was intended for someone with more patience than me.
todo.txt is (probably) fine if you are comfortable with priority ordered task management. I prefer date(/time) ordered management as it allows me to schedule tasks like buying birthday presents months or even years in advance and have them pop up when it's time to tackle them. My impression is that is difficult or impossible with todo.txt.
I often do that. Works file for stuff without time constraints. It could probably done even for tasks with start and/or finish dates, but I think it'd take more discipline than I could muster to make that work.
I find that the principle value of a computer based todo list is for infrequent tasks that I might otherwise forget -- change the oil in the cars, pay local taxes, run a quarterly blood pressure profile... Yeah, for a daily todo list, pencil and paper would likely be fine.
I failed to mention the obvious. EMACS Org mode has scheduling/task management features. As it happens, they aren't a great fit to my needs, but I think they will likely work better for others.
===
What does my Python task management software do?
There's a date ordered list of tasks
Each Task has:
an editable date with an (optional) date step (n days/weeks/months/years)
a title
an optional text field with unlimited amounts of free form text
Optional links to unix shell scripts
It has buttons/fields to step the date, edit the text, search titles or text, insert/delete/edit links. And it allows several different tasks lists to be in use at the same time. I use one for tasks and another for problems.
org-mode has all the capability one could desire -- outlining, integrated basic spreadsheets, embeddable links to all kinds of stuff. I actually do use it for some things.
**BUT**... EMACS key bindings are insane. There are good historical reasons for that. But nonetheless few people are capable of learning them, and even fewer actually do so.
**ALSO** EMACS has been around a long time -- longer than all common GUIs. As a result, many problems that are handled pretty consistently in browsers, Windows, KDE, etc are done differently in emacs. Things like search and replace. It's not that the EMACS solution is bad. It usually is fine once you figure it out. But it's different and needs to be learned or tweaked. Modifications exist for clipboard handling, text marking, and some other things that make EMACS work like other software. But there don't seem to be modifications for all capabilities -- search or search and replace for example. This may or may not drive the user nuts.
===
Two decades ago when I still used Windows, I tried every PIM I could find. I hated them all. I finally rolled my own solution in Python. I still use it. But I wouldn't really recommend that others do that. Two thirds of the code is convoluted and obtuse GUI handling that is enormously difficult to analyze and maintain. It turned out to be a great way to learn Python, experiment with procedural, object oriented and functional programming. And it taught me a great deal about why most modern software UIs suck so badly. But unless you are a lot smarter than I, rolling your own doesn't seem to be a quick path to a task management solution.
"except in cases of type 1 and clusters 1 and 2 in the article summary.....but yeah.."
I was going to say much the same thing. Even for Type 1, Carbohydrate management is probably advisable as it should make guessing proper insulin dosage a lot easier. The consequences of getting the dosage wrong can be serious or even fatal. And the $#@% blood glucose test strips don't seem to work anywhere near as well in practice as experimental results say they should.
Slashdot Mirror
"What's your proposed model for testing an autonomous car driving amidst normal traffic conditions that does not include actually having it drive among normal road traffic?"
The obvious answer is to install the hardware and software on vehicles that are driven by humans performing normal driving and not to enable the autonomous control of the vehicle until millions of miles are accumulated with no false negatives (failure to detect a problem) and minimal false positives. In practice, that's not going to work perfectly. But it's clearly not what Uber is doing.
"Companies like Google and Facebook make it much easier to correlate all the data."
The irony being that almost all that data is actually quite useless. Sooner or later -- maybe next week, maybe a decade from now, folks will figure that out and there will be a massive market reevaluation.
Talk about emperor's new clothes...
So, We will end up with a block market selling older permanently lockable phones to those who do NOT wish to share their secrets with the FBI, KGB. NSA, local law enforcement and the weird kid down the street? Here's your chance to make a killing folks. Get in early. Business Plan? When a decsion is needed, just ask yourself -- "What would Uber Do?"
Naw, probably the police just assessed the situation as they would for a human driver. Nighttime. Pedestrian in a place a pedestrian shouldn't have been. An unimpaired human driver might well have been unable to avoid an accident and very likely wouldn't have been held responsible. BUT. The damn car seemingly SHOULD have done better than it did. Disturbingly it never seemed to try to avoid the accident even though it might have been too late to do so by the time the pedestrian was recognized. THAT seems very bad.
Conclusion: Probably not really ready for prime time.
The vehicle has LIDAR, right? Pedestrian dressed in black combined with night might well be a problem for the human driver but it shouldn't be a problem for the hardware? If it is a problem for the vehicle, The same situation is likely going to be a problem in broad daylight with the pedestrian dressed head to toe in dayglow orange, No?
They don't much like you either
If Ayn Rand had had to deal with Comcast, she probably would have embraced Marxism.
FWIW, here's a link to the story in Plattsburgh's local newspaper which may provide a clearer explanation of the problem. http://www.pressrepublican.com... And yes, late December and early January did feature some unusually low temperatures -- around -17F(-27C) several nights presumably causing folks to run more electric heaters than usual more of the time.
I'm not a big fan of electric vehicles, And considering the price, I don't think anyone who much cares about cost per mile is likely to buy a Tesla. But we really shouldn't compare highway mileage for an ICE vehicle to EVs which I would assume will be used mostly for city driving and commuting applications where non-hybrid-ICE mileage tends to suck.
"The train is estimated to take 3 hours, compared to 5.5 hours by car."
Ten or Fifteen years ago, Mythbusters staged a race between a Bay Area starting point and an LA area destination with a pair driving a car and an individual taking a plane (nominally about an hour) then renting a car in LA. The racers arrived within a few minutes of one another. Their conclusion. Depending on luck, both take about the same time. My guess is that HSR would likely be about the same (fewer delays than air travel, but slower and won't sit at the airport for hours waiting for the fog to lift at LAX.
An even better example might be the Erie Canal which was built by New York State in the 1820s. NY built it after private capital refused to take the chance. The Erie Canal opened up settlement of mid-continent North America a couple of decades before the railroads got to be competetive. It cut travel time from the coast to the Great Lakes from a month to nine days. It turned places like Syracuse, Rochester and Buffalo into boom towns It cemented New York City's position as the nation's leading port. And it actually paid for itself.
That said, To me California' HSR project iooks like an (expensive) solution in search of a problem. I lived in California much of my life and I've also travelled a bit and lived overseas. My guess is that California probably needs a transportation system based on personal transportation, not -- with a few exceptions -- public transit. But maybe "cars" don't need need to be so big, intrusive, and generally obnoxious -- at least not in the Bay Area, LA Basin and San Diego. In the hinterland -- say Mono County -- I can't imagine what you'd replace them with.
Except that for political reasons, the newer, faster replacement is going to run through the distinctly not so scenic central valley rather than along the coast. And it's presumably going to cost a lot more than the Coast Daylight if they hope to ever recover construction costs. And there's a distinct -- once_you_get_to_X,_you_need_a_car_to_get_to_where_you_really_want_to_be problem in California's sprawling urban areas
I'm not against trains. I even ride them sometimes. But I think perhaps California needs clearer vision rather than a probably doomed attempt to imitate Europe, East Asia and the US East Coast corridor.
One thing that the slashdot description doesn't mention. Just having two transmitters in a room or building does not produce intermodulation products. You need some sort of non-linear "device" as well to combine the signals and emit the IM products -- in radio terms, a mixer. (That's not the same thing as an audio "mixer" BTW). I'm having a little trouble imagining what could act as a mixer and emit significant 7kHz audio signals. And wouldn't folks be able to find the mixer if not the transmitters just using their ears? 7KHz is well within the audio range for all but a few of us elderly folks.
In reality, the Cubans suggested that the US State Department to bring the FBI in to see if they could get to the bottom of the problem. (No, I don't know why the Cubans might think the FBI might know anything about sonic weapons).
Do the Cubans have the US offices and residences bugged? Of course they have the US offices and residences bugged. Why would they not? You can probably hear the plumbing talking to the air conditioners and every other electronic device if you listen real hard at night. Did they unleash some weird sonic weapon of the gringos? Even if they had such a device why would they use it on diplomats instead of on folks who are actually a threat?
"And why make it look like the wall,"
It'd be more fun to make it look like an archway going to an adjacent room or a garden and watch people walk into it.
====
But seriously, I imagine that people with more money than sense -- no shortage of THOSE -- will find uses for these. Put up a famous picture -- The Nightwatch or September Morn. And change it occasionally.
Heck, there might even be some useful things these things can do.
Can't access it? Security feature. It's undeniably secure.
Just wait until Ford or Toyota or VW inadvertently lets a certificate expire and a few hundred million vehicles glide silently to a stop ... wherever they may happen to be,
"I shouldn't need to eat up bandwidth and route traffic between Vancouver and Cupertino simply to copy data between two devices 10 inches apart."
I shouldn't be so quick to broadcast that sort of obvious nonsense to the world citizen. Just keep in mind that without profits there is no progress, and that progress is the only reason for your continued existence. Next, you'll probably be asking questions like where it is, exactly, that you are progressing toward.
It's not like the rules of addition and multiplication are evolving rapidly.
The todo.text format https://github.com/todotxt/tod... doesn't seem to have a simple date field. The closest I see is a FINISH START pair, each specified as a 10 character YYYY-MM-DD string. That's a lot of typing for a number that can be specified as YYMMDD. It's been a long time, but I'm pretty sure I looked at that and decided that the format was intended for someone with more patience than me.
todo.txt is (probably) fine if you are comfortable with priority ordered task management. I prefer date(/time) ordered management as it allows me to schedule tasks like buying birthday presents months or even years in advance and have them pop up when it's time to tackle them. My impression is that is difficult or impossible with todo.txt.
I often do that. Works file for stuff without time constraints. It could probably done even for tasks with start and/or finish dates, but I think it'd take more discipline than I could muster to make that work.
I find that the principle value of a computer based todo list is for infrequent tasks that I might otherwise forget -- change the oil in the cars, pay local taxes, run a quarterly blood pressure profile ... Yeah, for a daily todo list, pencil and paper would likely be fine.
I failed to mention the obvious. EMACS Org mode has scheduling/task management features. As it happens, they aren't a great fit to my needs, but I think they will likely work better for others.
===
What does my Python task management software do?
It has buttons/fields to step the date, edit the text, search titles or text, insert/delete/edit links. And it allows several different tasks lists to be in use at the same time. I use one for tasks and another for problems.
org-mode has all the capability one could desire -- outlining, integrated basic spreadsheets, embeddable links to all kinds of stuff. I actually do use it for some things.
**BUT** ... EMACS key bindings are insane. There are good historical reasons for that. But nonetheless few people are capable of learning them, and even fewer actually do so.
**ALSO** EMACS has been around a long time -- longer than all common GUIs. As a result, many problems that are handled pretty consistently in browsers, Windows, KDE, etc are done differently in emacs. Things like search and replace. It's not that the EMACS solution is bad. It usually is fine once you figure it out. But it's different and needs to be learned or tweaked. Modifications exist for clipboard handling, text marking, and some other things that make EMACS work like other software. But there don't seem to be modifications for all capabilities -- search or search and replace for example. This may or may not drive the user nuts.
===
Two decades ago when I still used Windows, I tried every PIM I could find. I hated them all. I finally rolled my own solution in Python. I still use it. But I wouldn't really recommend that others do that. Two thirds of the code is convoluted and obtuse GUI handling that is enormously difficult to analyze and maintain. It turned out to be a great way to learn Python, experiment with procedural, object oriented and functional programming. And it taught me a great deal about why most modern software UIs suck so badly. But unless you are a lot smarter than I, rolling your own doesn't seem to be a quick path to a task management solution.
"except in cases of type 1 and clusters 1 and 2 in the article summary.....but yeah.."
I was going to say much the same thing. Even for Type 1, Carbohydrate management is probably advisable as it should make guessing proper insulin dosage a lot easier. The consequences of getting the dosage wrong can be serious or even fatal. And the $#@% blood glucose test strips don't seem to work anywhere near as well in practice as experimental results say they should.