Over the past several months, we have seen a lot of interest and sales of our True View 6XX series LIDAR/camera (3D Imaging Systems, 3DIS®). We now have four different models in the market:
**We have sold out of our stock of miniVUX-2 and now offer only the newest miniVUX-3.
Performance amongst the 3 generations of RIEGL miniVUX sensors (1UAV, 2UAV and 3UAV) is essentially the same except for point density.
The maximum number of pulses that can be emitted per second (pulses per second, pps) into a 120° cross-track field of view (FOV) are:
Our new True View 515 (which uses a Hesai Pandar32 scanner) emits 213,333 outbound pulses per second into that same 120° FOV. So what makes the miniVUX so special as compared to these higher pulse rate sensors? Why should a customer pay so much more for a RIEGL-based product than lesser choices?
Well, I can give you a list of reasons, but let’s continue down this path of emitted pulses per second.
Sensors such as Velodyne, Quanergy, Hesai and the new Livox VIVA (of the DJI L1) are principally designed for the autonomous vehicle market. As such, they need to see fairly small objects in fairly short time periods (e.g. quick enough to avoid a collision). As such, they are more concerned with pulse density than noise or network accuracy.
My priority in automotive is “is there something in the roadway?” and “to within 10 cm or so, how far away is this object?” Since very fast pulse (short) collection is difficult, one can use independent, parallel lasers to increase the pulses emitted per second.
I created (in True View EVO, of course) an image of a single revolution of a PandarXT-32. If you can see the lines well enough, you will count 32, one for each beam. This means that I only have to fire each beam 20,000 times per second to achieve the aggregate 640,000 pulses per second.
Now this is a very clever way to get more pulses out there at a lower clock rate but look at the span of a single spin! I measured it to be about 50 m for this flight of 75 m above ground level (AGL). If I am flying at 5 m/s, it will be a full 10 seconds before the rear beam is at the location of the front beam.
This mixes a lot of temporally separated lines into a close geometric space. The result is increased vertical noise.
Since the miniVUX is a single beam sensor, the entire collection area can be devoted to that single outbound pulse. I took a (rather poor) photograph of the miniVUX collector mirror (see Figure 2).
As you can see, there is a lot of surface area to collect reflected pulse energy. This results in much higher sensitivity (signal) without increasing the noise (in the EE world, we call this increased Signal to Noise Ratio, SNR). Again, the result is a clean return pulse.
The result of these factors (and, of course a lot of other engineering detail) is a very clean (low noise) and accurate point cloud. Figure 3 is a hard surface (roadway) profile from a True View 640 (the 640 uses a miniVUX-3UAV).
The grid size in the profile view is 10 cm x 10 cm. I am not sure it is obvious in this figure but the noise level is just a few cm, peak-to-peak and this was a flight acquired at 75 meters AGL!
If you look closely at the left side of the profile, you will see a survey check point centered directly on the LIDAR profile line. This is a measure of network (“absolute”) accuracy. Again, it is spot on with no geometric post-processing.
There are a lot of adjectives for this technology; I just call it remarkable!! No wonder Departments of Transportation insist on this technology.