Tempe Police Release Video of Uber Accident

At 38 mph, a car is traveling about 56 feet/second. That means that the car traveled about 84 feet in that timeframe. VOL (Visual Aim on Left) headlamps are supposed to be aimed 2.1 inches below headlamp level at a distance of 25 feet [1]. If the headlamps on the vehicle were about 2 feet off the ground, then they should have been able to light up the roadway about 286 feet ahead.

Had the headlamps been aimed properly, then the driver could have seen the pedestrian about 5 seconds in advance. That would have given the driver about 1 second to react and 4 more seconds to slow down or change direction to avoid the collision.

[1] http://www.danielsternlighting.com/tech/aim/aim.html

I wouldn't blame a human for not seeing that person, but better should be expected of the tech.

However:

* there doesn't appear to be a single reflective device on the bike. For instance, the usual spoke-mounted reflectors that are stock equipment on even low-end bikes do not seem to be there.

* the woman seems completely oblivious to the car's approach. She doesn't react at all but keeps casually walking with the bike right until the moment of impact. She mustn't be looking in the direction of traffic at all and is mowed down completely by surprise, like someone sucker-punched in a bar. (Was this someone with disabilities? Visual or hearing impairment? Developmental?)

* I think here is the road where this took place: https://www.google.co.in/maps/@33.4351488,-111.9415554,3a,60... The scene of the accident is a little bit forward of here. Utterly not a place to be crossing at night in a way that is completely oblivious to the presence of vehicles and far away from the intersection. Note that this is a one-way double lane; there is only one direction in which to look out for cars.

This kind of badly behaved, suicidal pedestrian is a challenge to drivers even in daylight. However, you would expect precisely this sort of situation to be among the highest priority test cases for self-driving tech.

BTW: here is a shot of a sign forbidding pedestrians from crossing at almost that exact spot: https://www.google.co.in/maps/@33.4362927,-111.9424451,3a,15... She wouldn't have seen that one because she probably crossed the other lane already; she's coming from the median. How did this person live to 49?

1. They DO have cameras onboard with the exposure settings necessary to see the woman in the video, and have not released the video because it makes for bad PR.

2. They DO NOT have such cameras, in which case they should have their self-driving permit revoked because this failure mode is completely predictable to anybody working in this space. And anyway their LIDAR should've detected the person.

Edit to add: I don't really think they intentionally darken the image, but this probably isn't even the camera the car runs off of.

The ISO determines the sensitivity of the sensor in the camera to the photons hitting each bucket in the "sensor array". This sensitivity combined with frequency of the photon hits determines the data that makes up the actual image. If you could theoretically dump the RAW data from an image recorded for a different ISO for the same image, it would be different for each one.

You can't "boost" an image's ISO post-capture, because the sensor data is already captured. What you can do is increase the values of those pixels that are appreciably close to buy not equal to zero, but for which data exists, giving the impression of pulling out data from the shadows, but the limitations/complexities are too much to go into here. When you do so in practice, you do not have as much relative detail in those extremes.

What we interpret as "noise", typically in high ISO low light photography, is a real thing that the computer cannot perfectly compensate for, because it's a fundamental property of both the hardware, but more importantly, of background light/radiation and the probability/quantum nature of light...