In the Bee Lab, we are constantly trying to integrate more
and more technology so that we can be as precise (read: lazy) as possible. Normally
we have to go out and survey the field station on foot, going plant to plant
counting the flowers in the burning Summer sun. However, with the power of
technology we can send a drone in the air to take pictures of the flowers for
us. Then with some computer magic that we’re working on that I don’t even fully
understand we can make the computer count how many flowers are out there, it’s
genius! The one problem is, we do need to get out there and do the drone
flights over our research areas so that we can get those pictures to analyze.
Sounds easy, right? I thought so too. Turns out those little drones are a bit
more finicky than I would’ve thought.
The first week I started out using the same equipment that
was used last Summer, a DJI Phantom 3 Standard, the app Litchi, which allows us
to make flights with predetermined waypoints, and the remote controller. This
was generally fine for the first week, ignoring a few little hiccups while I
figured out the software and procedures (and some other confounding variables), and although
I occasionally forgot a few things back at the lab, it went fine. Week 2,
however, was a totally different story.
 |
| Image 1: One of the drone flights made with Litchi |
Near the end of the week, as I was doing my final map, I noticed the drone doing something a little odd in the air. The gimbal rig, which keeps the camera steady despite the rapidly fluctuating drone, was going absolutely crazy. It was spinning around in circles, turning the camera in all sorts of directions. As soon as I noticed, I landed the drone and turned it off, hoping the issue would be fixed with a quick power cycle. It wasn’t. When I turned it back on, same issue. I brought it back to the lab and checked the SD card. I found that the gimbal had actually gone crazy during all the flights I had done that week, rendering the photos useless for making maps. As a (hopefully) quick fix, I attempted to update the firmware, normally a fairly easy process. About half way through it stopped the update and made an excruciatingly loud beeping noise. I had to take it back to my room and put noise canceling headphones on to try to work on the drone so my labmates could still work in the room. Despite my best efforts, I couldn’t seem to get it fixed.
 |
| Image 2: A Phantom 3 Standard gimbal rig |
Talking to Professor Donaldson-Matasci, we determined that we needed the maps no matter what, since the maps were a critical part of the experiment. We would had to get a new drone. We had maps for week 1, and if we had maps for week 3, at least we could extrapolate any changes that may have happened during week 2. After some research, we decided to buy a DJI Phantom 4 Advanced, one of the newer models on the market from DJI. When it got here, I immediately got to work. The Phantom 4 was a lot easier to use than the 3, and had a much more powerful camera to boot. Through a good bit of testing, I was able to get it all set up and got my flights in for the week. Though I hoped against hope that they would, the problems didn’t stop here.
 |
| Image 3: The Phantom 4 Advanced |
Nearly every week I had some problem or other making my
maps. Sometimes the markers we had out in the field weren’t flipped up (see Michelle Lilly’s post on the markers for images) and I
didn’t know, so I would make a beautiful map of the field station where all of the markers that identify our plants were essentially invisible. Sometimes I would forget cables that I needed that I didn’t previously need
for the Phantom 3. Sometimes the wind was too high and I couldn't get stable pictures to make a map. It could be any number of things. One
thing that plagued me for a few days was storage space. I hadn’t really thought
about the fact that better quality pictures means the pictures will take up
more space on our SD cards. So, as our maps got bigger, we didn’t really have SD
cards on hand to match. As a result I had to go out several times each day, emptying the
SD card each time so I had enough space for the flight, until we got a few more
SD cards.
A little later on, as I went out for a flight I noticed that
my drone just wouldn’t take off, even though everything seemed like it was
ready to go. I took a long time out there trying to figure out exactly what was
going on, and eventually gave up for the day. Turns out a firmware update
changed the no fly zones for the area, and the warning area for the no fly
zones for Cable Airport and Brackett Field Airport had gotten a lot bigger and
covered the Bernard Field Station. Now I had to go online and get
individual authorizations to enter the warning area for each airfield, and
those authorizations last only 2 days. This means, with my terrible memory, I
often forgot to get the authorizations before I went out, so either I took 30
to 45 minutes to do it out in the field on my phone or gave up and decided to
do it a different day.
 |
| Image 4: The no fly zone unlock tool, the left marker is Brackett Field, the center is Harvey Mudd, and the right is Cable |
All of these little roadblocks didn’t completely stop the process, however. Other than week 2, I was able to get maps for all of our areas for every week we went out, even if it came close most weeks. It was a really fun Summer, flying around drones and working with all of my other labmates, and I will most definitely be spending some time writing procedures and troubleshooting tips so the next person has no problems at all (as if that will ever happen). In the end though, if I’ve learned anything, it’s that the lazy way can be a whole lot harder than I think.
Image Credits:
Image 1: Taken from the Litchi Mission Hub on the Bee Lab's account. Here is the tool itself:
Image 2: Taken from the DJI Phantom 3 store page: http://store.dji.com/product/phantom-3-standard
Image 3: Taken from the DJI Phantom 4 advertisement page: https://www.dji.com/phantom-4-adv
Image 4: Taken from the DJI no fly zone unlock tool page: http://www.dji.com/flysafe/geo-system/unlock
No comments:
Post a Comment