Tesla has revealed that the Cybertruck’s frunk is using a ‘learning algorithm,’ which explains some of the inconsistent results in finger-pinching tests.
Many people have had concerns about the Cybertruck’s sharp corners being a liability for users and pedestrians.
If you have the Cybertruck in your algorithm, you must have seen some of the countless videos of people cutting vegetables and fruits with the Cybertruck’s frunk.
Tesla recently released a new update with improvements to the frunk’s object detection system. The automaker wrote in the release notes:
“The powered frunk can better detect obstructions just before it finishes closing.”
Cybertruck owners who got the update and tested it have reported inconsistent results:
Wes Morrill, the Cybertruck’s lead engineer, commented on the video revealing that the Cybertruck frunk uses a learning algorithm for its pinch detection in the closing system:
Frunk pinch detection is a learning algorithm which will increase the closing force each time it’s cycled back to back without successfully latching.
This would explain why repeated tests produce inconsistent results.
Morrill added an example to explain the system:
Imagine there is a big bag inside the frunk it might also trigger the pinch detection. Then you might try closing it again, and again… exactly as you are doing in this video. The algorithm assumes that if you are repeatedly trying to close the frunk it’s because you, as the human in loop know better and believe it should close.
In short, Tesla wants the Cybertruck’s frunk to be able to close by itself through repeated strong closes if need be. Keep that in mind.
Electrek’s Take
Manufacturers have a responsibility to make things as safe as possible, but there’s a limit to making things dumb-proof. We heard similar concerns about the Model X’s Falcon Wing doors in the early days, and I don’t remember it being a problem.
However, it’s the first time I’ve heard of a learning algorithm being used in a frunk or trunk for that matter opening system.
That might be going a bit too far, but with improvements in AI, it could eventually work flawlessly.