Yeah. It had me thinking a few things... this implementation is a particularly dangerous one. Also, I found it interesting to learn that the majority of LIDAR on autonomous vehicles are using Frequency Modulated LIDAR? So when they say frequency modulated, are they talking about the wavelength itself, and not the pulse rate? So a wavelength could start out at say 910nm and then modulate down to 900nm, and the process repeats, just as FMCW radar does? I'm learning that I didn't know LIDAR as well as I previously thought...
Apparently you'll want to use a matching LIDAR-block filters on your lens if you know you'll be near these in the future.
Also, I found it interesting to learn that the majority of LIDAR on autonomous vehicles are using Frequency Modulated LIDAR? So when they say frequency modulated, are they talking about the wavelength itself, and not the pulse rate? So a wavelength could start out at say 910nm and then modulate down to 900nm, and the process repeats, just as FMCW radar does? I'm learning that I didn't know LIDAR as well as I previously thought...
Also from Ars: Continuous-wave frequency modulation lidar
The big challenge for these lidar-on-a-chip systems is that integrated circuits don't deal well with the peak power demands of conventional pulsed lidars. These lidars send out a brief, powerful burst of laser light and measure how long it takes for the pulse to return to the lidar unit.
"Aside from the signal processor, the only functional element that has not been demonstrated in photonic integrated circuit form is a high peak power oscillator or amplifier, because of the peak power handling limits of small waveguides," the authors write.
Until this problem is solved, the authors argue, lidar-on-a-chip systems will need to be built using a continuous-wave approach instead.
The authors point to a technology called frequency-modulated continuous-wave lidar as one promising direction. In the FMCW approach, the lidar generates continuous laser light with a steadily increasing frequency. This light is split into two beams, with one of the beams being sent out to hit the target object. When the beam bounces back, it's recombined with the second beam.
Because the beams traveled different distances and the frequency of the light was increasing over time, the two beams will have different frequencies when they're recombined. This will produce an interference pattern with a beat frequency that depends on how far the first beam traveled before it was recombined with the second. In principle, this should provide an extremely sensitive way to measure distances.
An additional advantage of the CWFM approach, Bowers and his colleagues argue, is that "up-down frequency ramps can be used to unambiguously distinguish range and velocity."
Here's what he means. CWFM lidars measure distance by measuring the frequency difference between a laser beam that bounced back from a distant object and a second laser beam that only traveled a short distance on the chip. If the distant object is stationary (relative to the lidar), then a larger frequency difference is a sign that an object is farther away.
But another possibility is a frequency change due to a doppler shift from the object moving relative to the lidar unit. A CWFM lidar can address this by taking each measurement twice: first with an increasing frequency, and then with a decreasing one. This switches the sign of the distance-frequency relationship, while the velocity-frequency relationship is the same for both measurements. Then it's a simple matter of algebra to figure out both distance and velocity simultaneously. (Aeva, a lidar startup we covered last month, seems to be based on a similar technology.)