A recent video has shown Tesla’s AI-based driver monitoring system in action, as it monitors the driver’s face using its inward-looking cabin cameras. As I’m sure you know, these cameras weren’t utilised by Tesla for a few years while it produced this system, promising that they would eventually be used to stop people from vandalising the cars, especially when being used as taxis when the cars are used in self-driving robotaxi network.
They were turned on a year ago in an OTA software update, where owners of the cars were asked to allow the cameras to take photos and videos of them while they used the car to “develop safety features and enhancements in the future”. Maybe customers volunteered, and now this work has come to fruition despite Tesla removing some volunteers from the program after it was confirmed that they weren’t paying enough attention while the system was operating.
Now, Tesla hacker Green has released a video of him using and analysing this feature.
As you can see from the videos, the top left shows what the car is ‘thinking’ and the probabilities of what it believes you to be doing such as using your phone, looking down, or looking ahead as you should be the majority of the time. Green has also released a video on YouTube of him trying to trick the system:
Very interesting stuff, and I’m looking forward to seeing how this is used in the future.