"the visible wavelengths required significantly lower laser exposure than near infrared wavelengths to disable subjects, though near infrared sources remain attractive given their cost and retina safety"
The paper repeatedly claims near infrared laser light to be "retina safe" (quotes theirs) which is an oversimplification. Infrared light of wavelength around 1 µm in sufficient powers is very dangerous for the retina, especially compared to visible light where the blink reflex can save damage by lasers up to a point. Telecom wavelengths around 1.5 µm happen to be safer because such light gets absorbed in the cornea before being focused onto the retina, but at sufficient powers or longer wavelengths (e.g. 3 µm) where the absorption depth in the cornea gets very short there can instead be damage to that part of the eye. It's always the combination of the wavelength, which determines the absorptivity in the different parts of the eye, and incident energy (pulse duration × power), that determines the safety; there are no innately safe wavelengths close to that part of the spectrum.
Maybe the best would be to mix a weak visible laser with a more powerful near-infrared laser, so that you make people blink, and your more powerful laser is less harmful? You could even trigger the visible "warning" laser 5-20ms before you fire the killer laser.
Some powerful laboratory IR lasers come with a built-in visible tracer laser. Mainly, it's useful for aligning all of the optics. With the IR laser enabled, I'm still wearing my safety goggles.
10 J/cm2 may be an adequate safety margin especially in near IR ranges where absorbtion occurs less in vulnerable inner eye structures.
There will definitely need to be some fail-safes to prevent beam from irridating one spot for more than necessary. Either beam movement or pulse length limits, or both.
Unrelated, I'm not sure why they need cameras. You already have a moving laser, couldn't you just scan the volume with it looking for insect wing beat reflections?
I guess you'd also need to look at possibilities that the laser light with the "safe" energy density of 10 J/cm² can't get focused by accident, e.g. by glasses or windows and whatnot, giving a higher and therefore potentially eye-damaging energy density.
> Unrelated, I'm not sure why they need cameras.
Machine vision stuff expects a pixel map as input. You'd have to implement some quick raster scanning step to build that, but I guess in principle it would be possible.
Yeah, laser safe with optics is a real concern. Again IR might be useful here, for some frequencies only exotic materials are transparent like quartz, salt, transparent aluminum.
Using the lasers you could measure insect wing beat frequency to avoid killing non target insects. I believe I read that in an older paper about insect laser fences, maybe these authors ignored it because it's patented
There are no safe wavelengths at all. Any frequency light at sufficient power levels will damage some part of the eye. Mechanisms may vary. Coherent light tends to be more dangerous but just looking at the sun will do damage and that's definitely not coherent light.
For every part of the spectrum there are safe limits.
The paper repeatedly claims near infrared laser light to be "retina safe" (quotes theirs) which is an oversimplification. Infrared light of wavelength around 1 µm in sufficient powers is very dangerous for the retina, especially compared to visible light where the blink reflex can save damage by lasers up to a point. Telecom wavelengths around 1.5 µm happen to be safer because such light gets absorbed in the cornea before being focused onto the retina, but at sufficient powers or longer wavelengths (e.g. 3 µm) where the absorption depth in the cornea gets very short there can instead be damage to that part of the eye. It's always the combination of the wavelength, which determines the absorptivity in the different parts of the eye, and incident energy (pulse duration × power), that determines the safety; there are no innately safe wavelengths close to that part of the spectrum.