General Discussion

FrodosPet

(5,169 posts) Wed Jul 6, 2016, 08:23 AM Jul 2016

Tesla Autopilot Fatality Shows Why Lidar And V2V Will Be Necessary For Autonomous Cars

http://www.forbes.com/sites/samabuelsamid/2016/07/01/first-tesla-autopilot-fatality-demonstrates-why-lidar-and-v2v-probably-will-be-necessary/

Sam Abuelsamid | JUL 1, 2016 @ 08:03 AM

The news Thursday of the first fatal accident involving a driver using the semi-autonomous Tesla Autopilot system highlights a number of important issues. First and foremost is that Autopilot does not magically transform a current Model S or X into a self-driving car. Far from it in fact. This crash also highlights that Tesla CEO Elon Musk was almost certainly wrong when he said last fall, “I don’t think you need Lidar. I think you can do this all with passive optical and then with maybe one forward radar.”

The person behind the wheel of a Tesla in Autopilot mode must remain alert to what is going on around them and be prepared to take over at any time. In its current incarnation, Autopilot should only be used on limited-access highways, and even then it will occasionally fail to detect lane markings and disengage itself. In urban driving, the forward-facing camera will read speed limit signs and display the posted limit in the instrument cluster and on the navigation map. It will not, however, read traffic signals or obey stop signs. The system also doesn’t detect pedestrians or cyclists.

~ snip ~

The more fundamental problem is the sensor suite that Tesla has installed on its current production models. Autopilot relies on a single monovision camera system from Mobileye, a forward-facing radar sensor, two side-facing radars in the back corners and 12 ultrasonic sensors. Unlike seemingly just about every other autonomous system from a major automaker or supplier, lidar, a technology that uses lasers to measure distance, is conspicuously absent.

~ snip ~

An optical camera suffers from many of the same issues as the human eye and is easily confused by many scenarios with less than ideal light, such as driving directly toward the sun at sunset. It’s also not very good at recognizing the edges of a snow-covered road. Or driving into heavy rain. Or fog. You get the picture, so to speak.

~ snip ~

1 replies

= new reply since forum marked as read

Highlight:

Tesla Autopilot Fatality Shows Why Lidar And V2V Will Be Necessary For Autonomous Cars (Original Post) FrodosPet Jul 2016 OP

Adding Some Statistical Perspective To Tesla Autopilot Safety Claims FrodosPet Jul 2016 #1

FrodosPet

(5,169 posts)

1. Adding Some Statistical Perspective To Tesla Autopilot Safety Claims

Reply to FrodosPet (Original post)

Wed Jul 6, 2016, 08:59 AM

Jul 2016

http://www.forbes.com/sites/samabuelsamid/2016/07/05/adding-some-statistical-perspective-to-tesla-autopilot-safety-claims/

Sam Abuelsamid

~ snip ~

As usual, we must ask, what question is this the answer too? Based on Tesla’s statements we can assume the 130 million miles is the total of all miles traveled in Autopilot mode by all Model S and X vehicles globally since the update was released in October 2015. If we assume that we must also assume there have been no fatal accidents in other parts of the world that we don’t know about yet. Given the amount of telemetry that Tesla collects from their vehicles let’s give them the benefit of the doubt on this one. So one fatality in 130 million miles stands for the moment.

~ snip ~

That leaves 22,276 vehicle occupants (drivers and passengers) that died. This latter set are probably the ones we should be comparing to Tesla’s one death in 130 million miles. By that count, one vehicle occupant died for every 135.8 million miles traveled with a human driver in control. Based on that statistic, humans are actually better drivers than computers. However, even that isn’t necessarily a valid comparison. While every death is a tragic loss, the reality is that one fatality in every 90+ million miles traveled is actually remarkably few.

~ snip ~

A data set of one in 130 million miles really isn’t enough to make a real statistical judgement. Those 130 million miles sound like a lot in isolation and does in fact amount to 87 times as many miles as Google GOOGL -0.79%’s test fleet has run around Silicon Valley and Austin, Texas. However, it’s only 0.0043% of the total traveled in 2014. Until we have a larger data set (not that I want more people to die, although they almost invariably will given enough miles traveled) it’s really too soon to make a judgement about whether we have actually made progress.

~ snip ~

I do believe that in time as automated driving technologies improve it will reduce the number of crashes, injuries, fatalities and the related societal costs. I’m just not convinced we have the data yet to prove anything one way or the other.

~ snip ~

Reply to this discussion