- Wed Jul 18, 2012 1:42 am
#147531
Basically, it's a time of flight detector that uses a continuous wave instead of a short pulse. It works by transmitting a specially modulated light beam and then comparing the transmitted signal to the received signal and figuring out how long it was in flight.
Let's run some numbers here. If the laser is modulated at 500 MHz, the wavelength of the modulated light is about 60 cm, or 30 cm round trip. With a phase detector, you can compare the transmitted light with the received light and measure the phase offset, measured in degrees from 0 to 360 (or -180 to 180, doesn't really matter). The phase measurement will repeat every wavelength, so a distance of 5cm will have the same phase offset as 35cm or 65cm or 95cm etc. If you can measure the phase to within 1 degree, you can measure to 30cm / 360 degrees = 0.833 mm. Using a second, much lower frequency you can figure out which 'bin' the measurement is in. 50 MHz with a round trip distance of 3m can be measured to 8.33 cm, enough to approximate the distance and differentiate between 5cm and 35cm at 500 MHz.
The circuit operates at RF frequencies (500 MHz) and so is relatively complex, but the result is almost an instantaneous measurement, allowing thousands of samples per second at sub-cm accuracy and sub-mm resolution. Also, as the detector only needs to be sensitive to a narrow band at 500 MHz, it is rather insensitive to external interference. Coupled with an optical filter for the laser wavelength, and it should possible to achieve excellent performance even with significant background illumination. For extended range, a more powerful laser diode can be used and the detector can be coupled through a telescope to improve the SNR. It would be possible to measure distances of hundreds of meters without losing resolution or accuracy. Long range accuracy is dependent on the absolute frequency stability of the driving oscillator and short range accuracy is dependent on the accuracy of the phase measurement. Think about it - 1000+ measurements per second with sub-cm resolution and accuracy!
Direct time of flight detectors require extremely high speed timers. To measure distances accurate to 1cm, you need clocks with a period shorter than 70 picoseconds - that's about 15 GHz! There are solutions that involve taking thousands of measurements per second at a lower frequency and then averaging, but these methods yield rather slow update rates. Also, the laser pulse needs to be very short and well-defined and it's quite difficult to reliably extract the laser pulse out of the background as it is so wideband. Generally the lasers used are multi-watt, pulsed lasers and the detectors are avalanche photodiode detectors. These detectors are extremely sensitive, biased at hundreds of volts and capable of providing a gain of 1000x or more. However, this is a very wideband gain and so the overall light hitting the detector must be very limited, requiring very narrow band optical filters and carefully designed enclosures. A continuous wave detector doesn't have as much gain internally, but it can be coupled through an RF amplifier chain with various frequency-selective stages and narrow bandpass filters, providing gain only to the band of interest and making the system more reliable.
The Parallax LRF calculates distance with triangulation, requiring processing a large picture pixel-by-pixel on the controller. The accuracy varies with distance as the angular size of the pixels (and therefore distance resolution) varies with distance. It is a relatively simple task to carry out on a general purpose controller, but there is a very well defined limit on how well it can scale as it becomes more and more difficult to measure the ever shrinking angle at long range. Moving the camera farther from the laser scales up the range - long range resolution degredation happens farther out and the minimum range is increased. Moving the camera closer does the opposite, decreasing the minimum range and moving the point of resolution degredation closer.
Think of it as x vs tan(x). Measuring the time of flight is directly measuring x, be it measuring the phase offset or flight time. The Parallax LRF measures tan(x), which has a very obvious limit at short and long range.
Alex Forencich