Hello,
In my search for a Lidar device for my Android telemetry system I have been testing both the Lidar Lite from PulsedLight (http://pulsedlight3d.com/) and the SF10/B from LightWare (http://www.lightware.co.za/shop/en/rangefinders-and-altimeters/32-sf10b.html).
I made a testsetup on a DJI Phantom 2 (Vision+) using both sensors and my GPS/altimeter module (MS5607). The two Lidars was interfaced thru I2C to my own board (interfaces the two Lidars to CAN bus). Data was sent on 433 MHz to the ground and stored on a SD card for later analysis.
Originally I got a Lidar Lite with a small problem, this was quickly swapped by the distributor (RobotShop) so the device tested should be pretty current (approx 1 month old).
The SF10/B was also acquired about a month ago or so.
In the test the Altimeter gave some erratic readings. This was because of light entering the sensor at times and very much from the wind from the props on the Phantom.
The sensors was probed around 8 to 10 times pr second, data was stored on the SD card around 9 times a second.
The Phantom was flown over a grassfield (very short grass) and over a field with approx 50 cm high crops, texture like high grass). There were no noticeable difference from the two surfaces on the sensors.
However, the Lidar Lite fails when reaching heights above approx 20 meters (advertised as capable of "at least 40 meters"), the SF10/B works to the advertised height (50 meter) without problems (worked to about 55 meters in my tests).
I have to be fair and say that the Lidar Lite was used with its default settings, if changing these could help I do not know.
I know that there is quite a difference in price betweenthe two sensors, and I also know (now) that the tress don't grow into the skies ;) )
Below is a image and a textfile with data for one of the flights
The blue line is altimeter, the red is Lidar Lite and green is SF10/B (Y axis is in centimeters)
Replies
I wrote:
All,
So as some may know I went forward with my 1st gen LiDAR units and wired them for interface with output rail 5&6. All the wiki parameters were set and a few test flights performed. So far the LiDAR unit is functioning but recent tests indicate it reliability to provide measurements outside 25meters has been problematic. See my post HYPERLINK "http://diydrones.com/forum/topics/copter-3-3-beta-testing?commentId=705844%3AComment%3A2002523&xg_source=activity" \t "_blank" here and HYPERLINK "http://diydrones.com/forum/topics/copter-3-3-beta-testing?x=1&id=705844%3ATopic%3A1964465&page=88" \l "comments" \t "_blank" here. I'm not the only one as Carsten Groen also observed ( HYPERLINK "http://diydrones.com/forum/topics/comparing-lidar-lite-and-sf10-b-in-the-real-world?commentId=705844%3AComment%3A2007127&xg_source=activity" \t "_blank" Comparing Lidar Lite and SF10/B in the "real world") similar issues with the LiDAR measurements dropping out and coming back in around the same range.
So I'm now asking the question... Is the measure faults a problem with the LiDAR units, software issues, or something else? My tests have been performed over grassy yard and asphalt. I still need to perform a vertical climb in Loiter to ascertain that this isn't from the motion of the copter and I should have something posted by end of this weekend. I feel that there is something not right and ask for you very smart folks to take a look at the information posted in the DIYDrones site and help me to understand what's going on. I'm hoping its just a software tweak that is needed to improve the LiDAR Lite, bust I just don't know at this time.
Thanks,
Doug W
Dennis wrote:
Hi Doug,
First let me thank you for all the research and work you've put into testing LIDAR-Lite. We appreciate your efforts and feedback!
To respond to the question you have about the discrepancies that you're seeing beyond 25 meters. The primary issue you will have with performance will be with the reflectivity of the terrain you're flying over. In testing of LIDAR-Lite we use a stationary target that is roughly 1m square and has a reflectivity of >80%. Under those conditions we are able to obtain readings in excess of 50m. However, when flying across dirt, grass or pavement you may only have reflectivity from these surfaces of around 20-30%. So the performance of the sensor will decrease as a result of the strength of the return signal. And if you throw in any angle, which effectively increases the distance, the return signal will also be weaker. So, in those cases readings of around 25-30m seems about right. There is an article about returned signal strength and effects of distance, target size and reflectivity in our documentation HYPERLINK "http://kb.pulsedlight3d.com/support/solutions/articles/5000566418-returned-signal-strength-effect-of-distance-target-size-aspect-and-reflectivity" \t "_blank" here .
Our signal processing technology does allow for adjustment of the length of the integration time for a distance measurement. The integration time is determined by the maximum number of acquisitions the sensor takes to develop a signal correlation peak in the return signal record which is then used to determine the distance. To understand this process better we have an operational overview HYPERLINK "http://kb.pulsedlight3d.com/support/solutions/articles/5000548647-operational-overview" \t "_blank" here. The default setting for the signal integration is 128. So it will take up to 128 acquisitions or, in the case of a strong return signal, it will stop before reaching that maximum setting.
The default acquisition time (count) is set to 128 as it is a good compromise between speed and performance. This is also the default setting that is used when operating in PWM mode. In I2C, HYPERLINK "http://kb.pulsedlight3d.com/support/solutions/articles/5000549552-detailed-register-descriptions-internal" \t "_blank" Control Register 2 , which controls the maximum number of acquisitions taken in a measurement, can be adjusted from 0-255. The lower setting decreases acquisition times and increases measurement speeds, but also decreases sensitivity. So it is only suitable for applications where the sensor is operating at short range or to high reflectivity targets. The higher setting increases sensitivity and also increases the time it may take to get a single measurement, so it is best used for applications where the target reflectivity and return signal are weak and the time for a single distance measurement is not critical. Currently, measurement times can range from a couple of milliseconds in the case of a strong return signal to a maximum of 20 milliseconds when using the maximum setting.
The default setting, 128 acquisitions, is used in PWM mode and is not adjustable. Ideally, using I2C would be the preferred method of connecting LIDAR-Lite to your drone. That would provide the greatest flexibility by allowing the sensor configuration to be changed on the fly (pun intended). Features like velocity and second return could be utilized as well as changing the system settings to accommodate different operating conditions.
Finally, it is probably good to mention that we are achieving the performance we do solely through our signal processing algorithms. Our transmitter is an inexpensive 5W laser which is being operated at under 1W of average power to achieve long life and the detector we use is a pin diode device, similar to what you find in the average TV remote, that costs approximately 30 cents. This is why we are able to offer the performance we do at the price we charge for the sensor. Competitive products are typically using higher power, more expensive lasers operating around 20W and very sensitive Avalanche Photodiode Detectors that in volume can cost $30 or more. The result is a much more costly system, but in some operating modes they may have advantages over our technology - today. LIDAR-Lite is just our first product.
I hope I've been able to shed some light on your question.
Cheers,
Dennis
that kind of confirms what we see I guess. I was "blinded" by the "at least 40 meter" detection range, guess I should have read that written with small letters ;)
Anyway, I did another test, logging the status register and signal level registers from the Lidar Lite and send it of to the company (Bob), maybe they can use the info for something.
The results are like the previous two test flights I did, the Lidar Lite stops around 21 meters (25 when going up and around 21/22 meters when going down). The SF10/B continues to work at 55 meters.
(oh, and the SF10/B was lowered I think $100 in price short time ago, now priced at $399,- :) )
I have attached the raw file from the flight, the screendump shows the heights measure (in cm) from altimeter, Lidar Lite and the SF10/B sensors. Again the altimeter is heavily offset because of the mounting and the props from the multirotor
20150604.TXT
I have been in contact with PulsedLight (Bob) and I'm getting yet another sensor to try out (this is the last test I'm doing, if no success with this sensor, the SF10/B will be selected for the project/product).
I will be back with the results and logfiles next week (weather permitting)
I received yet another sensor the other day (this is now the third, the first was very unstable, the second could only measure up to ca. 21 meters height).
It seems that this third sensor is a little better than the second, this one will go to around 30 meters.
Now, my main concern is how I/we can now if the sensors we buy will work to 20 meters (as the second I got), 30 meters (as the third one I got) or maybe only to 10 meters in worst case ??
Attached data from two flights with the 3rd sensor
3rd sensor - first flight.TXT
3rd sensor - second.TXT
Got a final answer from Pulsedlight, basically they say that the first two sensors I got was from the first production run and they have improved the product and firmware since that. The last sensor I got was from the second run and it should be more representative of what distance to achieve (but they also write that differences can occure between individual samples).
Also the 40 meters requires that you have a large reflective surface that reflects a significant portion of the light back to the sensor.
This I guess, does not exactly go for grass....
So, the SF10/B will be used here for the future, the data given for that works perfectly for grass and "farming" fields (even better than promised)
Not tooting anyone's horn here but I have two LiDAR Lite units installed on two different frames and I have seen it perform up to altitudes of 40 meters. See my post here and here. I found the LiDAR Lite price tag appealing for an ancillary sensor and it's thus far appears to be reliable. What I'm analyzing now is the range accuracy and intermittent zero measurements. Not sure but my last auto mission flew at near 40 meters and two distinct periods I observed quite a few drop outs. After reviewing mission replay and data logs, those dropouts occurred during 180 degree turns. I did not see any altitude loss or erratic behavior during those measurement dropouts.
I sure the SF10B ($499 USD) is a fine product but there price difference is what steered me to LiDAR Lite ($99) as a hobbist.
Hi Doug,
thanks for feedback! I totally agree that the pricetag on the Lidar Lite makes it very attractive, that also why I included that in the test hoping that it would be "sufficient" for the application (I would rather have to pay 1/5th of the price :)
In your second post, it seems that you also get dropouts at somewhere above 20 to 25 meters if I read the graph correctly ? (picture below). I my own tests over grass the Lidar Lite totally lost signal above 27 meters and often around 21 to 22 meters.
In my test, the Phantom was flying in straight lines (slowly) so there were no sudden changes of direction etc, and both sensors where aligned at 90 deg to the ground.
First link I shared was on my Y6B in the backyard over grass. The second link was my X8 over asphalt. Both appeared to show the same drop outs at some certain altitude. At this point, I would really like the owners of LiDAR Lite to chime in and explain these anomalies. Not that they are bad, but rather to make clear that there should be a optimal effective altitude vs maximum useable altitude for their LiDAR Lite. I'm beginning to believe that you are correct that signal degradation is occurring somewhere around 25-30 meters.
I agree Doug!
When looking at the first "dip" on the red curve on your graph, it seems more to be near (approx) 23 meters or so, just before you go up to 25 meters or so (two dips) above that it happens more often.
It worries me that this happens, there is still some room left up to the claimed "at least 40 meters". From my tests, anything more than 20 meters are not working. Now, I know my test was done over (very short cut) grass and it might be better over asphalt (your tests does not seem to validate that though...)
Anyway, lets see if there will be some official explanation to this, maybe there are some of the registers that can be tweaked etc.
I presented the data/problem to PulsedLight. They asked if I could repeat the test adding some more data to the logfile. I have now added the signal strength and status from the Lidar Lite, testflight will hopefully (weather permitting) be during the coming week. I will post results here too.