Tesla Motors is super advanced, but the company’s method of verifying that drivers are focused in autonomous driving mode is super primitive, essentially relying on monitoring the back torque that drivers occasionally enter on the steering wheel. On the one hand, Tesla has long been equipped with internal cameras for drivers, and critics have been asking tesla to use them to monitor drivers’ attention, but the company has refused to do so.
Now, Tesla hacker Green has discovered that the company has finally come to the fore for those who stuff oranges on the steering wheel, drink alcohol for parties or simply sleep, and start developing a driver monitoring system based on internal cameras.
When he examined Tesla’s car software, he found that its neural network was trying to class out the scenes seen by the internal camera into the following categories:
Tesla only began using internal cameras to collect in-car data a year ago, telling drivers to collect anonymous data before an accident to improve vehicle safety. Even in test mode, Tesla requires users to repeatedly confirm that the camera in the car is turned on. Here’s a description of Tesla’s privacy from the car’s video device:
“Help Tesla continue to develop safer vehicles by sharing camera data for your vehicle. This update will allow you to enable the built-in cockpit camera above the rear view mirror. If enabled, Tesla will automatically capture images and a short video clip before a collision or safety event to help engineers develop safety features and enhancements in the future. As always, you can adjust your data sharing preferences by tapging Control, Security and Security, Data Sharing, and Camera Analytics. “
It is not yet known when this feature will enter the production car ranks.