Andrew Tridgell has been testing laser rangefinder with ArduPlane, for more precise autolanding. Here's his report:
Testing the SF/02 laser rangefinder on a Pixhawk on my Meridian. Worked very nicely! Some results:
The baro showed a bit higher than the laser, but the laser was definitely much more accurate - the baro was affected quite a bit by the speed of the plane causing low pressure in the fuselage.
I connected it to the Pixhawk using the analog connection, on one of the two 3.3V ADC connectors. That gives very good resolution, and low noise.
Once PulsedLight's LIDAR-Lite changes to using a laser (as they plan to do) and fix the short range and long range issues (bad values at ranges below 1.5m and bad values at long ranges) I hope the two will have similar performance, and the I2C connection on the PulsedLight module will be a better fit. The LIDAR-Lite prototype I have now has a range of about 12m compared to 40m for the SF/02, and gives bad data at short and long ranges, so right now the SF/02 is better for the work I'm doing on landing, but longer term I'm hoping the LIDAR-Lite will be a great option
There is more discussion of this on the DIY Drones Dev list here.
Great! There can be small differences in these "anomalous" effects due to temperature and background light conditions. That might be why it was behaving previously but has now become more sensitive.
That's an exact description of the behavior I see, yes. I give it a try and let you know how it goes. Though the device definitely was not giving this anomaly when first installed.
Thanks for bringing this to our attention. Feedback is critical to improving our products and since we have such a low failure rate it becomes difficult to get enough information to find bugs.
In this case, the SF02/F is losing signal at long range and the errors are showing up only during this time. Once the ground comes back into range, the unit behaves correctly again. What we have found is that some of our units have been set with the receiver sensitivity a little too high during the manufacturing process. Under normal conditions, when a return signal is present, this doesn't have any affect because all the detection thresholds are well away from any noise. However, when the signal is lost during an out of range excursion, the unit attempts to locate a return signal by looking closer to the noise, and if this noise is too high it will occasionally produce a false positive result.
This issue is very easy to correct, but if you don't want to do it yourself then please just contact firstname.lastname@example.org and ask for help to get the unit modified or replaced as you wish.
Most people don't know that all our products are completely "hackable" in the sense that the internal settings can be changed through the USB port. The settings that are provided relate to many aspects of the performance, including the sensitivity of the receiver, and they are grouped into categories so that they can be more easily understood. There is a more detailed note here on how to hack the SF02 but I will give a brief description of how to change the receiver sensitivity below:
---------------------- Changing the receiver sensitivity of the SF02 --------------------
Using our LightWare Terminal application, plug the unit into a the USB port of a PC and press the "connect" icon. There should immediately be readings visible on the screen.
1. Press the SPACE bar to access the user menu. These are the standard settings for the unit and are the only settings normally visible.
2. Press the capital letter 'U' followed by 'Y' to unlock access to more menus that contain the factory settings.
3. Press the SPACE bar again the stop the unit measuring.
4. Use the arrow right key or '>' to move to the next menu.
5. Keep moving '>' right until the menu headed "*** APD settings ***" is reached.
6. Reduce the value of the first item, "1: APD breakdown voltage" by 1V from the value that is currently shown. Check that it was entered correctly and record the initial value in case you need to set it back later.
7. Keep moving '>' right until the menu headed "*** Signal settings ***" is reached.
8. Change the value of the first item "1: Differential threshold" from the default of 35 mV/s to 50mV/s. Check that it was entered correctly and record the initial value in case you need to set it back later.
9. Press the ESC key. The unit will start running again normally.
That's all there is to it. After the changes have been made there will be less noise in the system and the threshold of detection will be a little higher so you should see an immediate improvement in the behavior during the out of range condition. If there are still a few spike you can repeat the process with slightly more aggressive values.
We are reviewing our production procedures right now to eliminate this problem in future units. Special thanks to Tridge and Joseph for all their help.
Appreciate you taking the time! I am going to give it a shot. I will reach out to the folks at LightWare and find out what I can do to help!
Thanks as well to you Jacko for your notes!
Mine developed a lot of glitching. Randomly giving zero height but only momentarily and has never caused a false landing event. Thus far!
I really like this unit though - it is indispensable to get a decent landing with a large RPAS.
@Joseph, this is an issue with LightWare Lidars that I am currently working with LightWare on. I've seen it on the SF10/B and SF10/C. The current workaround is to use the menus to reduce the sensitivity of the Lidar a little.
I'll point the LightWare engineers at this thread so they can comment
Good point Jacko. I am posting a DF and a tlog. The tlog covers a few DF. I attached the biggest section of the DF.
Perhaps you could post the logs? I'm not really a wizz at this but may help.
I am having an issue with this. Many times it works perfectly. On occasion the height is read improperly and the plane flares. I know this as a message was added to tell me when the plane had entered the flare stage. The sonarrange numbers look very good, but the height is wrong. Maybe some issue caused from a bank/pitch angle calculation?
looks fantastic item. Do you encounter problems when banked over and the range finder working not looking vertically down at the surface? I am looking into this for my pixhawk UAV especially for auto land. so would be interested I you comments. Thanks.