Hacker News Reader

Comment by adamweld | Hacker News Reader

I think your response time calculations are wildly wrong.

The LIDAR sensor[1] being used here can pick up targets up to 120m away. I'm not sure about the RADAR or vision systems also in place, but even LIDAR alone should have been able to easily pick out the pedestrian with plenty of time to come to a full stop.

This is clearly poorly designed autonomous driving software, not a sensory deficiency.

View on HN

I watched it again and you might be right, there was likely time for something to happen, given the available reaction speed going 40mph on a highway. I'm curious what that something translates to and what affect it could have had on this situation (ie, swerving, slowing by x mph, etc).

Because you didn't do the math yourself, 40 miles per hour is just shy of 18 meters per second. So 120 meters is almost 6.7 seconds at that constant speed (more if you're slowing down). Start of video to collision is less than 5 seconds.

That should give quite a bit of time to slow down then or at least move slightly out of the way, even given the person may not have been detected exactly within the 5s+ of the video + factoring in computation speed + mechanical response times. Although again this is speculation as I'm not intimately familiar with how the objection detection works and what a bike w/ plastic bags may have looked liked crossing the road, plus what available options were at that speed and given the environment. Thanks.

The braking distance at 80 mph for a modern car is 320 feet (just under 100m) the car should have been able to come to a complete stop if the software had been using the LIDAR correctly.

This isn’t a highway. Mill Avenue is one of Tempe’s most pedestrian-trafficked major roads. And it has a slow 35mph speed limit, and is well lit (much better lit than the dash cam shows), because so many pedestrians cross it.

Reply on news.ycombinator.com