Griffin, Invisible Drone: Concept and Design
Not long after receiving our Insta360 One X camera, I became interested in the function of the 'invisible' selfie stick and how it the effect was achieved. In some applications, the inter-lens blind spot used to hide a pole mount was a burden (camera close to subject), though I began thinking of ways to exploit this void and also to begin to understand it's limits and nuances. For example, under certain motion or vibration, the footage is stricken with fairly severe 'jello', rendering it pretty useless, so ultimately this had to be considered in whatever we tried with the camera.
I'd mounted 360 cameras to drones in various ways and whilst I usually enjoy the result, you're always left with a decision to either show the drone in shot or constantly chase the viewing angle to avoid the drone coming into shot. I wanted to make a small drone capable of lifting a 360 camera and it being completely out of shot. Insta360 were due to release the One R camera at this time so we got an order in quickly. Fitting a complete aircraft power system and all in the One X blind spot was going to be virtually impossible as it is fairly shallow so the One R was the one to try.
Once the camera arrived (two months!!), I began measuring it and testing where blind spot extents were. I then had an envelope in which I had to accommodate all of the components necessary to make it fly.
1 - The inter-lens distance is 28mm. The two fish-eye lenses have a field of view just over 180 degrees.
2 - The blind spot between these two lenses was where I had to conceal the drone.
3 - The blind spot gradually tapers to become narrower further away from the lens.
4 - Stitch line
Mid-way through the design and research stage I thought it sensible to actually check that such a thing doesn't already exist and sure enough StanFPV has a modified three inch frame that accepts the GoPro Max called the Cine-Bird. That gave me confidence that the principle worked, though I'd have to go my own way for the One R. A few weeks later, he announced that the Cine-bird One R edition was available.
By this stage I was already a few iterations into the project and was convinced that my design would product the effect I intended. I had tested the stitching at numerous stages of the design process with physical prototypes and made changes based on strength, weight and blind spot clearance.
The entire frame was designed in TinkerCAD. Whilst I use Fusion and Meshmixer, I find that TinkerCAD's simple to use system aids creativity and places fewer barriers in the way of prototyping.
There were about a dozen CAD models/version before I finally settled on a fully printed design to print. I'd estimate I went through about 500g of plastic by the time I had a fully printed working prototype.
The final frame came out at a respectable 66g and included carbon fibre tube and strip glued to the PLA plastic frame with epoxy.
The intention was never to have this frame capable of taking a heavy landing or crash, though when put together, I was surprised at how solid it felt. With the lenses so exposed, the goal here is simply to not crash. Ever. Almost any impact or loss of control would mean a destroyed 360 module which really dictates the style of flying and what environment it's used in.
From the outset of the build I had intended to ditch the stock red Insta360 battery module saving a not insignificant 34g and taking the camera itself down to just 84g. This meant that the camera had to be powered from the main battery via a 5v regulator.
My intended battery for this set up was an 850MaH 4s, though the dimensions of the pack would not easily fit within the blind spot footprint. This problem took a few nights to find a solution to, mainly exacerbated by poor selection of batteries readily available in the UK and the state of shipping (especially international) at that particular time.
Ultimately I went with dual 650MaH 4s packs. These combined weighed more than the 850MaH single pack, though if you take away the weight of the One R battery, the additional mass is compensated for. At this stage it was all still theoretical, though I had made piece with having to run a heavier battery set-up than I wanted to.
Version 1 frame completed
Total weight 381g including camera and battery.
For testing purposes, most of the hardware was made up from spare components. The motors were specially selected for the performance and dimensions. I needed a motor that would sit within the blind spot of the camera. Emax 1404 ECO motors are also compatible and I have these to test, though from reviews, I opted the test the Xing motors first.
LDARC F4 FC iwth 20A 4in1 ESC and 200mW VTX - Link
Xing 1404 3800KV - Link
Caddx EOS 2 FPV Camera - Link
5V BEC - Link
Insta360 One R 360 core and 360 module
To say I was delighted with the first flight would be an understatement. Whilst there were some minor stitching issues as a result of the frame design, these were easily corrected and I was more relieved that there weren't any major vibration issues.
The initial flight highlighted a few areas of improvement, namely control of the camera once mounted in the aircraft and more importantly some of the frame still being visible in the stitched image.
Version two used the same hardware components. Changes to the frame include:
6mm carbon tube rather than 8mm.
Friction and strap fitment rather than 'pocket' design as per version 1
Weight shaving, new weight 378g (AUW)
Replaced carbon fibre strip battery tray with 3D printed cradle.
New motor mount design
Kept the Xing motors
Whilst the frame and new camera mount performed fine, I found this version had some jello issues which were not present in the first video. The lighting conditions were very different which could impact on the handling of vibration.
Flight time was in the five minute mark with steady flight which I was really pleased with. The aircraft is very stable and locked in with stock Betaflight 4.0 tune, though I didn't push the performance too much. Video range is modest, though absolutely usable.
In order to test out the vibration issues, I changed the position of the camera retaining strap to further forward (towards the lens mod). I believe the unbalanced weight of the two Insta360 modules was creating undesirable movement. The following test footage was taken in the evening in a fairly low light situation at the end of a group meet-up. The edit is a little crude, though demonstrates the cleaner footage.
With a better understanding of what works and what needs adapting, I am going to take this concept to the next level. Working within the blind spot of the camera whilst an interesting challenge, meant many compromises along the way. To address this, I want to de-case the 360 module to increase the inter-lens distance and allow for different mounting options. I had considered this at the start of the project, though didn't want to ruin a camera. Youtuber Strange Parts finally convinced me to look into this further as the 360 modele doesn't seem very complicated at all. What will be interesting will be how the change of distance affects the stitching. The distance I have in mind sits between the stock lens and the aerial edition split 360 module so I am anticipating some fun with the post production!
I'll end this with a shot of the failed prints we made along the way. Despite the careful measurement and spacing of parts in the design stage, sometimes a tenth of a mm can be important or a part may not be structurally sound enough and requires correction. Sometimes a part just simply not look good or doesn't look right either. However, sometimes the are just failures with the printing which is also a consideration throughout thee design phase.
All good fun though :)