3 ms·

Human fault aside, anybody care to speculate why Autopilot failed here? Was it a shortcoming of the radar, the vision system, or the fusion if the two? Is ultra

Human fault aside, anybody care to speculate why Autopilot failed here? Was it a shortcoming of the radar, the vision system, or the fusion if the two? Is ultrasound used at all for AP or auto emergency braking? I recently watched a great talk by Andrej Karpathy about Teslas transition away from radar towards vehicle speed estimation using only vision: https://youtu.be/gZ2SeiLjaEc https://youtu.be/gZ2SeiLjaEc I wonder if it would have prevented the tragic outcome here.

fouc 5y ago
I'm not sure what the limitations of radar are, but doesn't seem like it was being used in this crash considering the Tesla car just drove right into the truck.
- FriedPickles 5y ago
  Radar is a critical component of Autopilot and every other automatic cruise control system to my knowledge. So it was almost certainly active. A Tesla with failed radar system would not allow the driver to engage AP.
  - polynox 5y ago
    I don't believe it's a critical component of Autopilot. From the last public video release in 2019 [1], it seems like they use the stereo cameras to derive depth information, and use radar as a kind of ground-truth in informing the model. I would think the source-of-truth for what the Autopilot system thinks is the position of objects is the stereo camera, i.e. the radar is not part of the online flow, but is used offline to correct the model. See specifically at 2:20:22 [2]: > We use that radar to annotate what the vision is seeing, the bounding boxes that come out of the neural networks, so instead of human annotators telling you that this car, in this bounding box, is roughly 25 meters away, you can annotate that data much better using sensors. I take that to mean that the radar is used as part of supervised learning for updating the model offline, whereas just the cameras are used as input to the model. [1] https://youtu.be/Ucp0TTmvqOE?t=8254 https://youtu.be/Ucp0TTmvqOE?t=8254 [2] https://youtu.be/Ucp0TTmvqOE?t=8422 https://youtu.be/Ucp0TTmvqOE?t=8422
danso 5y ago
The NYT had an Carnegie Mellon professor examine the data and video and quotes his speculation: > In most of the video, the Tesla maintained a speed of 69 miles per hour, but just before impact it briefly increased to 70 m.p.h. then slowed in the final second, according to data from the car. > Mr. Rajkumar of Carnegie Mellon, who reviewed the video and data at the request of The Times, said Autopilot might have failed to brake for the Explorer because the Tesla’s cameras were facing the sun or were confused by the truck ahead of the Explorer. The Tesla was also equipped with a radar sensor, but it appears not to have helped. > “A radar would have detected the pickup truck, and it would have prevented the collision,” Mr. Rajkumar said in an email. “So the radar outputs were likely not being used.”
Isinlor 5y ago
My guess is the issue with sensor fusion. Radar throws a lot of false readings so there are many situations where it is ignored. For examples, readings of stationary objects are mostly false. My guess is that the speed difference was so big that also in this case Tesla decided that the radar reading was stationary and false.
- loopz 5y ago
  Stationary objects, also called obstacles. These systems can give people false sense of security and expectations, exactly at the wrong times. That kind of unreliability may be worse than having to drive and be attentive as a driver.
- FriedPickles 5y ago
  Hmm I think if a vehicle is moving even 5 mph it would be easily picked up by radar, and its only truly stationary things that are a radar "blind spot". I suspect its a vision or fusion problem--as the article points out, the lighting wasn't great. Why doesn't the car use ultrasound for AEB? Perhaps risk of false positives coupled with the fact that its range is so short it would only have time for limited braking anyway?

fouc 5y ago

I'm not sure what the limitations of radar are, but doesn't seem like it was being used in this crash considering the Tesla car just drove right into the truck.

FriedPickles 5y ago
Radar is a critical component of Autopilot and every other automatic cruise control system to my knowledge. So it was almost certainly active. A Tesla with failed radar system would not allow the driver to engage AP.
- polynox 5y ago
  I don't believe it's a critical component of Autopilot. From the last public video release in 2019 [1], it seems like they use the stereo cameras to derive depth information, and use radar as a kind of ground-truth in informing the model. I would think the source-of-truth for what the Autopilot system thinks is the position of objects is the stereo camera, i.e. the radar is not part of the online flow, but is used offline to correct the model. See specifically at 2:20:22 [2]: > We use that radar to annotate what the vision is seeing, the bounding boxes that come out of the neural networks, so instead of human annotators telling you that this car, in this bounding box, is roughly 25 meters away, you can annotate that data much better using sensors. I take that to mean that the radar is used as part of supervised learning for updating the model offline, whereas just the cameras are used as input to the model. [1] https://youtu.be/Ucp0TTmvqOE?t=8254 https://youtu.be/Ucp0TTmvqOE?t=8254 [2] https://youtu.be/Ucp0TTmvqOE?t=8422 https://youtu.be/Ucp0TTmvqOE?t=8422

danso 5y ago

The NYT had an Carnegie Mellon professor examine the data and video and quotes his speculation: > In most of the video, the Tesla maintained a speed of 69 miles per hour, but just before impact it briefly increased to 70 m.p.h. then slowed in the final second, according to data from the car. > Mr. Rajkumar of Carnegie Mellon, who reviewed the video and data at the request of The Times, said Autopilot might have failed to brake for the Explorer because the Tesla’s cameras were facing the sun or were confused by the truck ahead of the Explorer. The Tesla was also equipped with a radar sensor, but it appears not to have helped. > “A radar would have detected the pickup truck, and it would have prevented the collision,” Mr. Rajkumar said in an email. “So the radar outputs were likely not being used.”

Isinlor 5y ago

My guess is the issue with sensor fusion. Radar throws a lot of false readings so there are many situations where it is ignored. For examples, readings of stationary objects are mostly false. My guess is that the speed difference was so big that also in this case Tesla decided that the radar reading was stationary and false.

loopz 5y ago
Stationary objects, also called obstacles. These systems can give people false sense of security and expectations, exactly at the wrong times. That kind of unreliability may be worse than having to drive and be attentive as a driver.
FriedPickles 5y ago
Hmm I think if a vehicle is moving even 5 mph it would be easily picked up by radar, and its only truly stationary things that are a radar "blind spot". I suspect its a vision or fusion problem--as the article points out, the lighting wasn't great. Why doesn't the car use ultrasound for AEB? Perhaps risk of false positives coupled with the fact that its range is so short it would only have time for limited braking anyway?