I had already asked that question, so now that you have we should hopefully get some theories.
I agree, but that's not comparing apples to apples. A human wouldn't be driving while looking at at video. Neither was the car for that matter.
I'm 66. I tend to repeat things.I had already asked that question, so now that you have we should hopefully get some theories.
Not much. Just thinking out loud. If these cars are supposed to be able to navigate and know what's going on around them, then why shouldn't they be able to change lanes to avoid an obstacle.
Not sure how you think that has any significance to my post. I never claimed that pedestrians don't get hit. I simply claimed that the vast overwhelming majority of drivers have faced a very similar situation at some time and avoided a collision. That it was not in any way shape or form an extraordinary situation.
"The more likely explanation is that the Uber vehicle's dashcam was poorly configured for nighttime recording, and so the video gives a misleading impression of how bright the scene was and how much warning the driver had."So people in Tempe are starting to see something wrong with that Uber video. It's too dark.
Here is a nighttime pic from a resident at the same location:
Here is the article:
https://arstechnica.com/cars/2018/0...victim-came-from-the-shadows-dont-believe-it/
Yes looks that way. Maybe the city replaced them after the accident.I'd put my money on Google's effort's leading the pack since they've been at it so long and would have had a much more significant brain pool.
I notice a significant difference between the images:
My guess is that it's just due to the camera in the car not being configured well for nighttime video. It's probably because the car isn't relying on the video alone to make determinations but all other sensors. My guess is that something failed on the car or something is simply improperly configured. The more I read, the less I trust Uber to be doing these types of tests. They're a company well known to cut corners and violate laws to get ahead. I have far more trust in someone like Tesla and Google doing this than I do Uber.
We will take a generation to be able to process that the much lower risk (higher than zero though) of driverless tech is easily a better call. We have such a bias toward thinking things things are so much better (or worse) than they really are. I heard a guy I really respect on a podcast candidly admit that even though he knows logically that it’s ridiculous, he wants driverless tech to have zero risk because he truly buys into it. Said he knows it’s just silly but it’s just easier to accept 35-40k lives lost a year to people driving badly than to accept 2-4k lives lost to “robots”.
The article I read stated radar/lidar as one of the technologies used by these cars.
The Uber test vehicle, which had a human safety driver who didn’t appear to be watching the road based on a video released by the Tempe Police Department, was loaded with sensors, including LiDAR and radar that can see pedestrians and objects in dark or in light. And even if the LiDAR, which shoots out pulsed laser beams to create 3D, 360-degree images of a vehicle’s surroundings, failed, the car’s radar should have detected the metal bicycle that Herzberg was pushing as she crossed North Mill Avenue.
Clearly need to improve that tech then. Tech wise may not be so hard. Difficult part is programming how to respond to this stuff. Someone is jaywalking and car can avoid it by hitting another one, what’s the call? I say run over the jay walker (not saying this was the choice here, just that we can do 0s and 1s raise than decide the accident values ahead of time). I will say I think this case is one where people may be asking more of the machine than a person, which we will do in the long run.
I linked another article up above on the autonomous car company used here by Uber not having the best record and were not granted the rights to test their cars in San Francisco. I have a strong feeling that this will point to issues with their prototype. Also not sure how all those sensors, LiDAR and radar were being used to control (stop) the car.
I’m not bothering to even watch too closely yet. There will be errors. Like Dr Owens says on Stranger Things suggests (this is paraphrase) we make lots of errors in advancing science. This tech will be clunky then not clunky then hit that critical mass where tech will advance so fast to ready state that we won’t even really catch it. So I won’t overfocus on the trip, I’ll wait on the destination. Well I will watch to see what I want to buy as part of the winning tech.
I do agree with you about reducing accidents/injuries/deaths should be the goal and no way should society expect it to be zero.
Well then they will pay. If they were smart they would leave ALL settings to the designer/manufacturer. There is no value in messing with them. In fact regulations should prohibit that. It’s an area in which we want experts being the ones who tinker, and even they will get it wrong some.
Well aside from the obvious reasons of less accidents and the abilitity to get on I10 and just relax on your trip. Hal makes a point. The idea is there will be a fleet of cars constantly moving around the city and you just call up one and it takes you to your destination. No need to have your own car. Had a few too many on Saturday night, get the ride home is still a great use of this.
As long as we are still allowed to drive our own cars. Because I often like to go where other people aren't going.
Doubt we will run into a Red Barchetta situation. Can’t imagine not being allowed to own compliant self driving autos even if many don’t. It may not make sense financially but I’m sure some will want to. Bigger issue is having non autonomous cars out with a majority fleet of autonomous cars. As tech improves, one way it will do so is by having cars communicate with each other, which should increase safety a lot. Adding unpredictable human driving aspects could make that a lot harde to do. Not saying we won’t have driver cars out there with autonomous, just pointing out the stuff I think about. I think 30 years from now the world won’t be recognizable compared to today when it comes to personal travel.
This is a great point. It’s interesting to me how safety appears to have gone full circle in some ways. 40 years ago (and more recently) we didn’t even use seatbelts and people routinely drove after (or while) drinking. We’ve reduced death and injuries in so many ways. But now we have so many more distractions that I usually assume someone near me is texting or playing on their phone while driving. Earlier in the thread we discussed the fact the “minder” in the Uber wasn’t paying attention but the reality is many actually driving aren’t paying attention. The future autonomous autos will take care of our devolution from paying attention to this dangerous activity.
