The race - terrytaylorbonn/auxdrone GitHub Wiki
24.0819 (0712) (WIP)
The ZiptieAI drone project focuses on what you need for the mission (building autonomous AI drones). And ZiptieAI subscribes to the "The SpaceX school of project management" (not Boeing). You should be in a race to win.
To win this race (against your own limimted amount of time to invest int this project)
- dont waste your time
- dont let the "experts" slow you down
Dont waste your time
I was working on a project in early 2024 where they wanted a fiber optic drone. No radio signals sent... everything over fiber. I wanted to meet the "customer" in Ukraine, those who would be using this to find out their requirements. But that was verboten, project leaders never really made me a part of the team.
Interesing project. I dont know if that project ever became useful on the Ukrainian battlefield, but later ran into a westerner working with Ukrainian drone builders. Someone with experience with the customer. His assessment: Fiber is a total waste of time.
I wasted a lot of time when I started this project (maybe not wasted... I did learn something). The screenshot below shows some of the topics that (mainly ROS and simulation) that I thought were important (because thats what I read). I now see them as a waste of time (at least if you have the same goals as I did). At least when you are starting out. The simulation is interesting, but its too complex, and provides little help for flying drones. ROS may be good for robots, but its too complex for most drone builders (drone != robot).
Dont let the experts slow you down
About a year ago the software dev team for the project I was working on had a meeting about Git Copilot (the new AI SW dev assistant). To me Copilot was merely the latest in a long line of SW dev tools developed over many decades (the previous generations of tools included compilers, high-level languages, development environments, "intelligent" auto-completion assistants, etc).
But for the team Copilot was primarily a threat to their livelihoods, not unlike what high tech textile machinery was 200 years ago to the Luddites.
It was probably the same a generation ago when the first AI web search algorithms appeared. These algorithms were a threat to the livehoods of the experts in many businesses and professions. For example, bookstore chains were not happy when Amazon.com started making books vastly more accessible to mankind (with online search and purchasing). Librarians were also undoubtedly not happy when intelligent online search algorithms made their expertise in library card catalogs useless (or when Amazon became the new libary of convenience for the vast majority of the population).
AI search (as implemented in online discussion forums, Stackoverflow, Youtube videos, Amazon.com, etc) in a somewhat similar way has disrupted drone development. Indeed, I was able to bypass many of the (very time consuming) traditional aspects of getting up to speed in drone dev (such as joining an RC club so that some revered expert might help me with endless questions and problems) and instead get started immediately building flight-tested drones from the comfort of my kitchen workshop (I even made the maiden test "flights" on my kitchen work table!).