The "terminator" in the movie 30years back had a similar device. Must be fun at parties - in 5 years from now you can mark the singles in the room automatically in the overlay display of your contact lens (face detection, facebook in the background), or - in china for example - the social credit score of everyone in the street as they walk by. I hope someone listens to Albert when he asks for 'responsable design'.
I haven't read the full details but my first thought was how do they make sure they know which was is 'up'? As presumably the lens can rotate within the eye, and you don't want to project image to be the wrong way 'up'!
I thought such lens have a little "thicker structure" in the middle that makes lens get aligned with every blink of the eye lid? There are also haloween lens that "make you look diabolic" where the print on the lens gets aligned with the blinking of the eye.
The chip architecture looks vaguely familiar: A 1.5-V 550-uW 176x144 Autonomous CMOS Active Pixel Image Sensor by Cho, Krymski and Fossum DOI: 10.1109/TED.2002.806475 (0.35um CMOS 3T APS, 5um pixel, cap SAR ADC) I guess this is lost in the fog of history unless one does a literature search. So, 20 years later the power is reduced by less than 10x. Not bad but not amazing. And, they claim that column-parallel ADC is much higher power than serial ADC for the same total throughput. Hmmm.
~25x improvement actually You need to take into account the higher pixel rate Their claim is correct since this one runs comparatively slow. For advanced nodes the ADC achieves optimal power efficiency at high speed while column parallel would need to be duty cycled which is imperfect... also takes more space.
I guess the fascinating part of this project is the application (and less the sensor technology). But I am wondering if they have really a working prototype (that can be put into the eye)? How do they solve the optical system for the camera and projector (depth)? And the power supply?
What kind of image sensor did they use?
ReplyDeleteThe "terminator" in the movie 30years back had a similar device. Must be fun at parties - in 5 years from now you can mark the singles in the room automatically in the overlay display of your contact lens (face detection, facebook in the background), or - in china for example - the social credit score of everyone in the street as they walk by. I hope someone listens to Albert when he asks for 'responsable design'.
ReplyDeleteI haven't read the full details but my first thought was how do they make sure they know which was is 'up'? As presumably the lens can rotate within the eye, and you don't want to project image to be the wrong way 'up'!
ReplyDeleteUsually contact lenses with astigmatism correction are heavier on one side to make shure that side is down.
DeleteHowever, in this case it may be unnessecary because the camera and projector rotate together.
Ah, good point! I hadn't thought it though. Also interesting to know about the astigmatism correction lenses
DeleteI thought such lens have a little "thicker structure" in the middle that makes lens get aligned with every blink of the eye lid? There are also haloween lens that "make you look diabolic" where the print on the lens gets aligned with the blinking of the eye.
DeleteThe chip architecture looks vaguely familiar:
ReplyDeleteA 1.5-V 550-uW 176x144 Autonomous CMOS Active Pixel Image Sensor by Cho, Krymski and Fossum DOI: 10.1109/TED.2002.806475 (0.35um CMOS 3T APS, 5um pixel, cap SAR ADC) I guess this is lost in the fog of history unless one does a literature search.
So, 20 years later the power is reduced by less than 10x. Not bad but not amazing. And, they claim that column-parallel ADC is much higher power than serial ADC for the same total throughput. Hmmm.
~25x improvement actually
DeleteYou need to take into account the higher pixel rate
Their claim is correct since this one runs comparatively slow. For advanced nodes the ADC achieves optimal power efficiency at high speed while column parallel would need to be duty cycled which is imperfect... also takes more space.
I guess the fascinating part of this project is the application (and less the sensor technology). But I am wondering if they have really a working prototype (that can be put into the eye)? How do they solve the optical system for the camera and projector (depth)? And the power supply?
ReplyDelete