Mark Rober on his second version of the glitter bomb, a revenge prank targeting package thieves and video-capturing their reactions, added some creative improvements to his original design. A fair number of his design strategies can lend some inspiration here, as well as other enhancements.
He uses four of “the world’s finest [cellphone] cameras” (in his case he uses the entire cellphones themselves, one on each lateral side of the box, though we’d just use the cameras of course). Cellphone camera technology has become increasingly sharp in resolution through the demand for its development, so we’d have extensive options utilising tiny cameras if we end up wanting a very lightweight drone. For proper scouting out and observation of habitats we are gathering from, we may want to use at least 5 if not more cameras to capture around and below the drone for maximum viewing.
As an alternative to having multiple surrounding cameras and to avoid blind spots (besides the drone itself), we could alternatively use a small 360° camera that can be mounted to a drone, which is done in examples here and here with a regular front camera for more traditional steering. This might be particularly useful for observing the habitat around the sample or for observing natural habitats themselves if scouting for samples. We would want the camera mounted below the drone in order to capture the full landscape, possibly suspended fairly low down or not directly below to minimise the amount of above camera space blocked out by the drone.
Size constraints and considerations:
If we do go for the small 360° camera option we might then want to consider making the drone itself as small or narrow as aeronautically possible to maximise video capturing, which would be easier to do with only one camera. The size of the drone and thus the power required to fly a long distance would affect the requirements of the battery.
The size in turn would then be constrained depending on the size of samples we are collecting.
For example a small sample of soil, leaves or insects that that can fit in a small container or test tube or a small liquid sample that can potentially be captured by an eyedropper.
If we do want to collect larger samples, it would be ideal to add this functionality later as a feature. Starting with a lightweight drone for collecting smaller samples would be easy to optimise for sustained flights and quicker to build a working demo.
It would require lighter constraints - we would require less power, a smaller lighter specimen container, less work in order to balance the sample being carried (as carrying a test tube of liquid or a small container of leaves or sand is far easier than transporting a living rodent for example) and we can also use lightweight robotics materials requiring less PSI of strength to gather samples as we would then be scooping a small teaspoon or two of sand, snipping small leaves or using our aforementioned eyedropper to draw small samples instead of sawing off larger twigs or potentially more complex lifting and balancing care in transporting something as large as a captured rodent.
It is also worth noting even a small sample drone may still need to handle more weight and balancing than a surveillance drone and we would still potentially be treading new ground by adding in the robotics for gathering, but minimizing size and strength constraints would still be easier in early iterations.
One of the benefits of his use of four cellphones was the inbuilt technology for using high location accuracy in case of GPS failure. While the Bluetooth and wifi detection functions may be less useful for navigation out in nature, using cellphone towers would help with another means to pinpoint accurate location than GPS trilateration, if, for example, only two GPS satellites are reachable. Thus even if this happens, also transmitting GPS coordinates and being tracked, the location at which it went down could then be identified in order to do any necessary retrieval. This is not to say we would need to use cellphones themselves (again except possibly in the prototype phase), but that we could consider using the same technology of hybridising GPS and also include cellphone signal in a separate lightweight phone or custom transmitting device as a backup/complementary signal device.
One of the most promising and new alternatives to GPS is how NASA and Near Earth Autonomy are working on breaking drones’ dependency on GPS in the case of its failure in order to enable autonomous drone navigation technology.
Autonomous navigation as a later feature
In terms of the abovementioned autonomous navigation technology itself, from a developmental standpoint, this has the potential for massive scope creep for reasons listed below, and so it makes sense to exclude it from the initial draft versions, where our concern would be getting the core functionality of stable and prolonged flight, specimen collection robotics and coordination, camera work and reliable navigation in low-signal areas working.
The reasons would be that autonomous drone navigation is still bleeding-edge technology, which also means potentially less supported, more prone to errors and more expensive and we would want to minimise the number of features while building out the core technology and getting a working prototype as soon as possible - this would be crucial in order to demonstrate our concept to potential investors and secure funding.
Livestream and Live Backup
The live stream in this case also includes live backup of footage in case the unit is destroyed or loses signal/battery or is otherwise unrecoverable so that no footage is lost with the drone. In this case, it would be ideal to deviate from Mark Rober’s method of backup, he uses cellphone data given his phones only start recording when the box is opened and only record a short video, which preserves both data and battery, but this obviously would not work for us in recording continuous drone footage so in the interests of avoiding using expensive and potentially intermittent mobile data for video backups directly from the drone, it would make sense to instead back up the stream from the controlling device (or a companion device) as it receives the drone’s video stream via First-Person View Technology, or FPV, which uses radio signal instead.
This technology can currently extend to a range five miles in regular drones and thousands of miles in the case of more specialised drones like military drones, making it feasible for the types of distances we may be considering for sample gathering although its specialised nature would add to its cost and effort to implement. However, with many companies like Amazon pushing long-distance drone technology the related flying and monitoring technologies will likely become ubiquitous and thus soon more affordable, supported and adaptable for our purposes.