AI drones - terrytaylorbonn/auxdrone GitHub Wiki

25.0126 (24.1231) Doc URLs Stack URLs Lab notes (Gdrive), Git

This page summarizes my AI drone experience.

  • Sep '23: AI/drones in Kyiv.
  • Spring '24: Test flying drones in my kitchen.
  • Summer '24: Successful flight test of AI drone.
  • Late summer '24: The autonomous seeker drone.
  • Sep '24: Back to APIs.

Sep '23: AI/drones in Kyiv

  • The customer product I was working on was reorganized (my contract was not renewed).
  • I then studied AI by doing examples.
  • While in Kyiv, I started exploring drone simulation (SITL).

Spring '24: Test flying drones in my kitchen.

After returning the US, I started to work on AI on 3 main platforms:

  • Linux
  • Jetson Nano
  • PI4/5

I also started to explore the quadcopter ecosystem (simulation, construction, flying, firmware, etc). My goal was to quickly build a carrier for the Nano or PI4/5. I spent a few months learning how to build and fly Pixhawk and FPV drones (the following show my initial flight tests; peforming these in my kitchen was my own idea).

image image

Summer '24: Successful flight test of AI drone

In the summer of 2024, just before a trip to Kyiv, I got basic object recog with jetson nano and flight control with pymavlink tested in my local park in the USA. In this video the drone is parked a few meters above ground and

  • I walk in front of camera
  • The Jetson Nano (AI computer) recognizes me as human
  • The Nano uses Pymavlink to command the copter to land

image

A simple demo, but it took quite a while to get it all running with a startup that used no computer connection. Expanding on this would have been relatively easy.

Late summer '24: The autonomous seeker drone

While in Kyiv I started research for building my own autonomous seeker drone (focused most efforts on Kalman filters). My goal was to find a (volunteer) job in Ukraine with a drone builder (Western NGO or Ukrainian).

Sep '24: Back to APIs

After returning to the US, I decided to focus on API/API-doc dev because of the following:

  • Better job prospects.
  • Higher productivity. I was spending the majority of my time with AI drones not exploring AI and autonomous fly (as originally planned), but rather hacking around to find solutions to low level drone hardware/firmware problems. Originally the drone was supposed to be just the carrier platform for the AI, but instead had become itself the focus of my efforts.
  • I needed to move to a better AI companion computer than the Jetson Nano. The logical step up was the Orin Nano. However, the web reviews for the Orin Nano were not good (a lot of posts talked in detail about how the complexity of configuring the operating system was even worse than the Jetson Nano).