131–140 of 445 posts
Doesn't this implie that autonomous driving won't be solved until we're able to have "true" personnal assistant ?
The message in driving are simpler than with a regular voice communication, but the objects and concepts visually present seem just as diverse.
Earlier quoted context omitted.
It shouldn't matter at all how dark the roads were, the vehicle was equipped with LIDAR.
It's easy for Uber to point to this video to say: "Look know dark it was! If it was you driving, you'd be in the same problem we are in!" Look how much time has passed to learn of the fact that it was Uber's sloppy programming at fault in this tragedy. The good news for Uber is that this comes out far after the event when people are in general less likely to care, or even remember the details of the accident.
What's there to "sloppily" program here? Most likely, somebody set an arbitrary threshold constant 0.05 too low. The data that threshold was working with will still have been provided by the machine learning system, which is opaque to humans.
We'll have to face it that if we want self-driving cars, there isn't some magic procedure we can follow to not have accidents due to false classifications.
I'm curious what degree of "cultural context" is built into automated cars - I feel that there has to be a lot of locale-specific techniques for driving. For example, in NYC, there's nothing at all uncommon about a pedestrian beginning to cross the street in front of you and approaching the area of your lane where you will shortly be. Pedestrians typically then stop about a foot away from your lane, let you drive pas…
NPR had a story about how self-driving cars in Pittsburgh wouldn't be adhering to the "Pittsburgh-left": "Even harder for autonomous cars to master are local quirks; things like "the Pittsburgh left." That's a custom unique to the city that allows the first driver trying to make a left turn to do so before oncoming cars pass through an intersection. It's one thing that Emily Duff Bartel says Uber's fleet of self-driv…
Earlier quoted context omitted.
People don't want more highway patrol, red light cameras, and traffic enforcement and tend to see it as a waste of resources. It also disproportionately screws lower income people and breeds contempt for police - people feel they are being persecuted when there are non-victimless crimes happening every day. I can't find a source for this, but I imagine they are not ever making net profit on tickets when a trooper mak…
Fines can increase without hiring more people or buying more cars. Fines could be made proportional to income, at least one other country has already done that. Why would they need to make a net profit? That standard isn't applied to anything else the police do. We already have traffic enforcement as well as many other police services paid via taxes. The argument that it costs too much to increase enforcement in any…
Fair point.
>Why would they need to make a net profit?
At the end of the day the resources either come from somewhere else or it costs more taxpayer money. Yes, I agree it would improve safety.
>The argument that it costs too much to increase enforcement in any way doesn't really hold much water.
It does though because it is the only solution and the fact that is costs too much is the main reason it isn't done already.
>So how would you make roads safer?
That isn't my job and it isn't a pressing political issue in most places. If autonomous cars exacerbate the problem, they should lobby the state and local government to spend more on traffic enforcement to enable them to operate safely.
I'm curious what degree of "cultural context" is built into automated cars - I feel that there has to be a lot of locale-specific techniques for driving. For example, in NYC, there's nothing at all uncommon about a pedestrian beginning to cross the street in front of you and approaching the area of your lane where you will shortly be. Pedestrians typically then stop about a foot away from your lane, let you drive pas…
NPR had a story about how self-driving cars in Pittsburgh wouldn't be adhering to the "Pittsburgh-left": "Even harder for autonomous cars to master are local quirks; things like "the Pittsburgh left." That's a custom unique to the city that allows the first driver trying to make a left turn to do so before oncoming cars pass through an intersection. It's one thing that Emily Duff Bartel says Uber's fleet of self-driv…
I'm curious what degree of "cultural context" is built into automated cars - I feel that there has to be a lot of locale-specific techniques for driving. For example, in NYC, there's nothing at all uncommon about a pedestrian beginning to cross the street in front of you and approaching the area of your lane where you will shortly be. Pedestrians typically then stop about a foot away from your lane, let you drive pas…
NPR had a story about how self-driving cars in Pittsburgh wouldn't be adhering to the "Pittsburgh-left": "Even harder for autonomous cars to master are local quirks; things like "the Pittsburgh left." That's a custom unique to the city that allows the first driver trying to make a left turn to do so before oncoming cars pass through an intersection. It's one thing that Emily Duff Bartel says Uber's fleet of self-driv…
Edit: oh sorry misunderstood what pittsburg left meant. I was describing situation when both cars turn left. There are 2 options: cube path, that is paths cross, cars pass one by one, or 2 arc paths not intersecting.
The question I want to know is: Did Uber doctor the camera video to make the road seem darker than it actually was? If you recall, there was initially a doubt that Uber was at fault because the streets appeared extremely dark, but I think they were buying time to delay the outage at incompetent Uber. The police initially didn't fault Uber based on this video.
That was a bad call on the part of the police. You're only allowed to drive as fast as conditions permit. If it's dark enough that you can't avoid an obstacle by the time your headlights illuminate it, you need to slow down .
Earlier quoted context omitted.
The law as written and the law in practice, are different matters. NYC would not function with actual adherence to the law. There is never a break in pedestrians long enough for a car to turn in a fully legal fashion at most intersections, which is why human drivers wait for the largest mass of the built-up crowd to go and then inch their way through the stream of those just arriving at the intersection (with pedestr…
Imagine how much safer NYC would be if the cars and pedestrians were isolated. A dedicated car layer with an extra lane on each side (at current level) and a new pedestrian layer built above the cars with physical barriers preventing falling to the level beneath.
It used to be that cars, trains (and horses) traveled at just over walking pace, pedestrian crossings (and jaywalking) didn't exist, pedestrians crossed the street anywhere.
https://youtu.be/aohXOpKtns0?t=4m4s
Though I guess today's traffic densities wouldn't allow for slowly meandering through traffic.
Earlier quoted context omitted.
I can recall thinking the video was very bad quality. To the point where I was questioning if the video was from some dashcam they installed instead of the self driving car tech.
The video was from a cheap dashcam, not any of the sensors the car uses. The LIDAR would have seen it clearly. So would your eyes, which have wider dynamic range than any but the best cameras.
Earlier quoted context omitted.
I was proposing it as a way to validate engagement by the testing teams, not the mass market.
Like I said, these are especially hard to get right for frequent users, presumably your testing teams would log a lot of hours on these - just like train drivers, which have had this exact same problem.