Linux 6.2 Adding Driver For Google's ChromeOS Human Presence Sensor

[phoronix]

Phoronix: Linux 6.2 Adding Driver For Google's ChromeOS Human Presence Sensor

Set to be merged in the Linux 6.2 is a new driver for the ChromeOS Human Presence Sensor "HPS" used for detecting when one or more humans are in front of the Chromebook...

https://www.phoronix.com/news/Linux-6.2-Chrome-HPS

[cynic]

nice feature!

I already see an useful usecase: soon, in order to watch a YouTube video you'll need to be in front of the chromebook during the whole advertising or your video won't start.

[skeevy420]

And to know when the women have left the room.

This part is for the men only.

[skeevy420]

Regardless of anti-ad users, that would be a pretty cool feature

Realistically I imagine this would be useful to lock Chromebooks being used in public locations if you have to get up or move away from the laptop at some point.

Or CPU/GPU throttling. No need to ramp speeds up if the person walked away. There are a lot of interesting things that can be done with HPD..human presence detection. Tying it to a video game pause function could be useful...as could mute or increase volume (varies by context). Another one, related to what you suggested, is turning the camera/biometric sensors back on when you get back to your Chromebook to enable automatic unlocking if it detects you as you.

Personally, I'd prefer a CPD sensor...cat presence detection. I don't like it when my cats type up 49 pages of text in Catlish or Kittenese or manage to find the F12 key to run previous terminal commands.

IMHO, the only real problem with this is that it's from Google/Alphabet because nobody trusts them anymore. That's what happens when you drop "Do No Evil". People assume you now mean "No, Do Evil".

[coder]

Has anyone looked at how it actually works?

These sorts of things can enhance security and provide other conveniences, but if they fail to work reliably for even 0.1% of users, they can present a major obstacle.

For instance, some "liveness detection" techniques, in face recognition systems, rely on pulse-detection. However, there are some artificial hearts which use a simple impeller, and they result in the owner not having a pulse.

[coder]

Or CPU/GPU throttling. No need to ramp speeds up if the person walked away.

Oh great, so if I start a build and then go get some coffee, I come back to find it's only 25% as far along as I expected? Not cool.

turning the camera/biometric sensors back on when you get back to your Chromebook to enable automatic unlocking if it detects you as you.

That's not presence detection. What presence detection could do is automatically turn on the camera and run a full-fat face recognition engine to authenticate you. This would save energy vs. having the face recognition engine continuously run whenever the screen is locked.

That's fine, as long as it can also be manually triggered (i.e. if HPD doesn't work for reasons like what I mentioned) and if other authentication methods are supported.

Personally, I'd prefer a CPD sensor...cat presence detection. I don't like it when my cats type up 49 pages of text in Catlish or Kittenese or manage to find the F12 key to run previous terminal commands.

That's a cool app idea. You could probably run some opensource object classifier, like YOLO, and use it to lock the keyboard if a cat is detected close enough to the camera. By "probably", the only thing I'm not sure about is if a normal Chromebook app can trigger keyboard-locking.

IMHO, the only real problem with this is that it's from Google/Alphabet because nobody trusts them anymore.

The article says even the firmware is open source. That might not always remain true, but for now it's a good move.

As mentioned above, the main risk is that it's used in some way detrimental to the user's interests.

[skeevy420]

Oh great, so if I start a build and then go get some coffee, I come back to find it's only 25% as far along as I expected? Not cool.

I believe I said context depending...but...who TF compiles stuff on a dual core Chromebook? I know this is Phoronix, but jeez.

That's not presence detection. What presence detection could do is automatically turn on the camera and run a full-fat face recognition engine to authenticate you. This would save energy vs. having the face recognition engine continuously run whenever the screen is locked.

That's fine, as long as it can also be manually triggered (i.e. if HPD doesn't work for reasons like what I mentioned) and if other authentication methods are supported.

I think we got our wires crossed because that's what I was describing. The low power HPD chip detects a person so it turns on the higher power drawing biometrics, camera, etc.

That's a cool app idea. You could probably run some opensource object classifier, like YOLO, and use it to lock the keyboard if a cat is detected close enough to the camera. By "probably", the only thing I'm not sure about is if a normal Chromebook app can trigger keyboard-locking.

My fluffy bastard has already ass-keyed gibberish and refreshed the browser once on me today.

The article says even the firmware is open source. That might not always remain true, but for now it's a good move.

Honestly, I just don't trust anything from Google these days. While this is open source, it's Apache licensed so all Google has to do keep up with "doevil.patch" behind the scenes.

As mentioned above, the main risk is that it's used in some way detrimental to the user's interests.

So it's OOMDv2