Laser range finders can be very effective in detecting even small obstacles at long distances from a moving UAV. To cover large areas the laser beam is usually scanned by direct drive or using mirrors. As an alternative to this approach, we present a multi-beam solution that has no moving parts and is therefore smaller and lighter than a scanning system. This new configuration, called the SF33, is ideal for "forward looking" obstacle detection along the flight path.
Starting with the electronics from our SF30 laser module, we added two additional laser sensors into the existing housing to create the 3-beam pattern that you can see in the picture above. These beams have 5 degrees of angular separation resulting in a fan shape that spreads out with distance. The beams can be arranged to give either a horizontal or a vertical pattern depending on how the housing is oriented, or they can be aimed 45 degrees downwards to give a "look ahead" indication of both obstacles and rising ground.
The SF33 is scorchingly fast so we are able to measure by cycling through each beam in succession to get independent distance results. In our prototype we set the cycling rate to 3000 times per second so that each beam could detect power lines from a UAV moving at 60 kph.
The "elastic band test" has become the benchmark for high speed obstacle detection, so we subjected the multi-beam laser to this indignity. We were expecting the elastic band to be detected three times as it passed through each of the three beams.
In the 'scope image below, the orange trace gives the distance as an analog voltage and the blue trace is a digital alarm signal that goes low when an obstacle gets too close. You can clearly see three sets of alarm events, one from each beam, as the elastic band shoots past.
We also need to consider that the beams are cycling 3000 times per second, so as the elastic band passes through the first beam, the second and third beams are still checking for other potential obstacles. This effect manifests as "gaps" in the alarm signal as can be seen in the zoomed-in traces of the first set of alarm events.
In this image, the three beams cycled 12 times as the elastic band traveled through the first beam. The alarm signal shows a 1:3 mark (low) space (high) ratio as beams 2 and 3 interleave their measurements with beam 1. A similar picture was produced by the elastic band as it traveled through the other beams.
The elastic band test clearly demonstrates that the SF33 multi-beam laser has adequate speed to detect obstacles with a high relative velocity and we have confirmed a useable range of around 50m when measuring to larger obstacles.
At this stage of development we can still make hardware and software changes. I know that a number of forum members have experimented with obstacle detection using lasers, so any suggestions on how to integrate multiple distance readings and alarm events into a meaningful communication message to the flight controller would be most welcome.
Special thanks to Jordi Munoz for his support and technical input on this project.
Comments
Thanks Eric, we will be showing this and other new technology on the InterDrone Expo in September.
maybe someday we can use this on our Drone
great work
great work
That sounds fine - we'll include the "combined beam" as an option for both the analog and serial outputs.
I since the simplest and best solution is to combine all three beams (in horizontal patern) for just one analog signal output with a 50meters range.
Positioning the two sf33 facing at ? degrees for covering a maximun front area.
The final issue is to using this two laser modules exactly like the maxbotics sonar with the ardurover code:
http://rover.ardupilot.com/wiki/common-optional-hardware/common-ran...
Hi Jean-Luc,
Thank you for the feedback - this sounds like a great project.
The optical end of the SF33 is splash resistant so it's better to have a cut-out in your enclosure and mount the unit against the inside front face using silicon sealer or gasket material with the lenses protruding through the cutout - there is a small flange around the SF33 for doing this. Using this method, the lenses are unobstructed and you will get the best signal strength with the lowest probability of false readings.
How would you like to process the muti-beam part of the data? We can combine all three beams so that the SF33 behaves as though it has a single, wide beam or/and we can make results from individual beams available using requests through the serial port. Any thoughts on what would work best for your application?
Good job LD!
I expect to use two modules in the ardurover mode like classical acoustic sonar use, one to command right motor and one for the left, with the analog output.
I need to put these modules in a waterproof enclosure, but what is the restriction regarding the enclosure window? glass or type of plastic, distance, thickness and so... or may be your front optic enclosure is already waterproof IP66/67 (just for splash is ok, no need immersion)?
Regards,
Safety is always a primary concern and so all our products are made to meet international and FDA standards of eye safety, namely Class 1M or Class 1. In the case of the the SF33 multi-beam unit, we have taken additional precautions by making each beam safe in its own right, then also making the combination of the beams safe.
In the talk about these small systems, I haven't noticed anything about the potential for eye damage. In Canada, to fly with any laser system on a UAV, it has to be ok'd by a health and safety board. Is there any problem looking down the barrel of one of these?
-
1
-
2
of 2 Next