News for hackers. Stuff that matters.
Re:Uber’s self-driving car could not detect pedestrians outs... by mzs on Wednesday November 06, 2019 @03:39PM (#21466970)
No it shouldn't have applied the brakes at that point. There are lots of vehicles around in most cases. If the car braked every time it spotted another vehicle it would almost never move. Only take corrective action if the paths look like they will intersect instead.
Re:Uber’s self-driving car could not detect pedestrians outs... by chungus_khan on Wednesday November 06, 2019 @03:50PM (#21467079)
At five seconds from impact it absolutely should have. And if it couldn't tell that it was that close to the "vehicle" that's even more reason it shouldn't be anywhere near the road in this state. The system should be able to behave cautiously in situations where input is erratic. Good human drivers don't go full speed ahead when they can't see.
Re:Uber’s self-driving car could not detect pedestrians outs... by mzs on Wednesday November 06, 2019 @04:13PM (#21467366)
It took over four seconds for the system to decide swerving was not sufficient based on path prediction. The improvements made after this report would have initiated braking seconds earlier.

-5.6s 44mph Classification: Vehicle - by radar Path prediction: None; not on the path of the SUV

- Radar makes the first detection of the pedestrian and estimates its speed.

...

-1.5s 44mph Classification: Unknown - by lidar Path prediction: Static; partially on the path of the SUV

- Lidar detects an unknown object; since this is a changed classification, and an unknown object, it lacks tracking history and is not assigned a goal. ADS predicts the object’s path as static. - Although the detected object is partially in the SUV’s lane of travel, the ADS generates a motion plan around the object (maneuver to the right of the object); this motion plan remains valid—avoiding the object—for the next two data points.

-1.2s 43mph Classification: Bicycle - by lidar Path prediction: The travel lane of the SUV; fully on the path of the SUV

- Lidar detects a bicycle; although this is a changed classification and without a tracking history, it was assigned a goal. ADS predicts the bicycle to be on the path of the SUV. - The ADS motion plan—generated 300 msec earlier—for steering around the bicycle was no longer possible; as such, this situation becomes hazardous. - Action suppression begins.

* The vehicle started decelerating due to the approaching intersection, where the pre-planned route includes a right turn at Curry Road. The deceleration plan was generated 3.6 seconds before impact

https://dms.ntsb.gov/public/62500-62999/62978/629713.pdf

Re:Uber’s self-driving car could not detect pedestrians outs... by jacquesm on Wednesday November 06, 2019 @04:21PM (#21467476)
> If the car braked every time it spotted another vehicle it would almost never move.

In this particular case I would be perfectly ok with that. If you can't operate a vehicle safely then coming to a stop or failing to get moving is fine.

Re:Uber’s self-driving car could not detect pedestrians outs... by mzs on Wednesday November 06, 2019 @04:27PM (#21467544)
When I drive there are things I see which I am not sure what they are but I don't care since they are stationary over there and I am going that a way over here instead. I don't stop to take a good look before continuing.
Re:Uber’s self-driving car could not detect pedestrians outs... by jacquesm on Wednesday November 06, 2019 @04:32PM (#21467606)
If they're in your path and you can't tell the difference is what we're talking about here. And that difference is very important because the dynamic characteristics of a pedestrian, bicycle, car or other object should factor into your model of their future trajectory and speed.
Re:Uber’s self-driving car could not detect pedestrians outs... by Piskvorrr on Thursday November 07, 2019 @07:16AM (#21472479)
"There's something straight ahead of me, unclear what. Let's just ram into it, full throttle!" If that's your thought process, please cease driving immediately and permanently.
Re:Uber’s self-driving car could not detect pedestrians outs... by bobobooey on Wednesday November 06, 2019 @04:58PM (#21467930)
If the car can't detect what's in front of it, then it most definitely should brake. This is up to the engineers to solve this issue, not for a vehicle to continue into a dangerous situation blindly.
Re:Uber’s self-driving car could not detect pedestrians outs... by smileysteve on Wednesday November 06, 2019 @08:30PM (#21469573)
> There are lots of vehicles around in most cases.

In this case, we know there weren't many vehicles around.

This very scenario is a great example where I'd want a car to stop if it saw a deer or even a dog or armadillo.

> If the car braked every time it spotted another vehicle it would almost never move.

In defensive driving it's often taught that you are supposed to slow whenever you're approaching another vehicle and don't know what it's doing. You're supposed to exercise caution at intersections, and definitely supposed to exercise caution when passing, being passed, there are things in the road or other people or vehicles on the side of the road.