Anyone who has used or worked with drones knows that they are interesting platforms on which to mount a camera. Some, from the a cheap Hubsan for around $30 on Amazon, to the much more capable DJI Inspire for around 60 times that price (Amazon), come with them already mounted on them. One thing most, or all as far as I know, don’t have is a spherical camera, allowing you to take “video” for use in virtual reality environments.
This type of recording, done from a drone or otherwise, allows playback to happen later as if you were an observer in a vehicle. You can look around, but can’t control where you are going. 360.place, a startup in the Ybor City district of Tampa, Florida, has been pioneering what can be done with this technology, concentrating on unique views of businesses. As their expertise focuses on software and video production, they asked me to help with the challenge of mounting a heavy (around 2 pounds) GoPro Omni VR rig to their DJI Inspire quadcopter.
Naturally, I was thrilled to take on this challenge, and after about a month of work designing, building, and testing my rig, I delivered a newly-augmented drone to 360.place that they could use to take their VR to the skies!
The obvious solution to taking this footage would be to run some sort of pipe from the center of gravity of the quadcopter a few feet down to the camera. This has been done before, but I didn’t see a good way to mount it in the middle without modification. Being a customer’s drone, I much preferred to do something that could easily be removed, and the linked design removed the built-in camera, which I wanted to avoid if possible.
After agonizing over what to do for a week or so and not really coming up with a good solution, the idea finally hit me while driving. As with many good ideas, it seemed obvious after it came into my head, but I could forgo attachment to the bottom altogether and run rods to the sturdy and flat carbon fiber sections attached to the motor pods, forming a stable platform. The only hitch in this plan was that that motors retract for flight, so I had to employ pivoting joints at the end of each rod.
Design and Build
Once I had the basic design in my head, it was time to actually build it. Although I perhaps could have started cutting and ordering parts immediately, I decided to instead do things properly, and modeled it in 3D using Onshape. This went smoothly, including (if you’re familiar with this kind of CAD) the slightly challenging mate where the four rods come together to restrict the movement of the box on the bottom.
One challenge I hadn’t anticipated was how to accurately model the triangular portion seen on the yellow OMNI mockup. After experimenting with various plane configurations, I found that using a chamfer on each edge took care of this surprisingly-tricky angle.
Once things were modeled, I put together a bill of materials of what I would need for the project, featuring three of my favorite parts sources: Servocity, Amazon, and McMaster-Carr. This (brainstorm, design, order, build) has been my basic machine design procedure for years, and seems to give me good results. Once I had my parts, I generated a few dimensional drawings for what I needed to cut and went to work!
The only parts I really had to machine were custom inserts for the carbon fiber rods and a carrier for the camera itself. The inserts were glued on with epoxy, and allowed the rods to be attached to rod ends on the camera side and inline ball joint linkages on the other. This meant that when the DJI Inspire raised its landing gear, the camera mount could flex along with it.
Although it’s tempting to skip careful planning and design, you have to “pay” for this in one way or another, perhaps in extra parts that you end up ordering (though I generally order a few extras just in case), or the fact that you have to remake something several times. Or if you need to modify something later, it’s much simpler if you have accurate drawings. Not that I never make mistakes, but it’s a lot easier to correct them virtually than in real life!
Initial Test, Initial Failure
Speaking of mistakes, after quite a bit of work it was finally time to test the rig. I set up a custom landing pad for it made out of PVC pipe and wood, and attached the 2 pound weight that I made using wood and weighted inside with spare drill bits. This would allow us to test its capacity for lifting the GoPro Omni without risking a $5000-or-so VR camera rig. As seen in the video here, things didn’t go as planned.
It seems that when PJ, 360’s drone pilot, lifted off of the pad, the Inspire tried to compensate for whatever drag was attached to it. Since this weight was around 3 feet away from the original drone’s center of gravity, this caused it to swing back and forth like a pendulum and eventually go into a sort of “death spiral,” or perhaps “death wave” would be more accurate.
360.place decided not to go with a “different” engineer, and we decided to cut the distance between the center of the quadcopter’s mass and the weight roughly in half. After changing a few things around design-wise, as well as several propellers, it was time to test it again.
For the next test, we decided to start off slowly, testing it with the rig and no camera first, then a cheaper (and much lighter) camera from Samsung. Finally, we tested it with weight representing the Omni, which the quadcopter was able to lift and maneuver with successfully. We tested it one more time after this, with the actual GoPro Omni as shown in the above pictures. After some final words of instruction, I said “goodbye” to my creation, leaving PJ to use it himself in an actual shooting situation.
The shots below are taken using this rig. If you’re not familiar with this type of video, use the arrows to change your view and look around. PJ tells me it’s quite a bit more stable than what they tried before, although I’m sure there is still room for improvement. To be fair, what’s seen below was shot using the smaller non-GoPro rig, as I’m told that it will take some time to produce the GoPro footage. Also, one of the cameras in the Omni ran out of power on the shoot seen here, so this is what I have for now.
This was a great project, and I was happy to take on something innovative for a new customer. If you have something you need built, quadcopter or otherwise, I’d love to hear from you. Check out my “Jeremy Cook Consulting” page to find out more!