15.1 AI Kamikadze - terrytaylorbonn/auxdrone GitHub Wiki

24.1224

6 months ago, before trip to kyiv, got basic object recog with jetson nano and flight control with pymavlink tested on myself in my local park in usa. In this video the drone is parked a few meters above ground and

  • i walk around to be in front of camera
  • nano recognizes me as human
  • nano uses pymavlink to tell copter to land

Primitive, but took quite a bit to get it all running with a startup that used no computer connection. Expanding on this would have been relativekly easy.


I was thinking about trying to create an autonomous Kamikadze drone, or at least research it. But did not want to invest my own time in this without a team for support. In any case, my thinking on this:

  • Fiber optic is great, but has limitations.
  • Get the fiber optic drone to a location, where it sits and scans. Sits because moving around would exhaust the fiber.
  • Then if the connection is lost or the drone spots a target,
    • you fly manually to hit the target wiht teh fiber connection
    • if connection is lost drone goes into kamikazde mode.
      • It drops the extra weigth (and center-of-gravity + aerodynamic annoyances caused by the fiber spool)
      • Uses AI object recognition to track target
      • uses Kalman to predict future location of object (only needed for fast targets, like other drones)
      • used PID to steer itself optimally to the predicted intercept point

About the last 2 bullets above:

  • Kalman tutorial (WIP). The Kalman filter predicts the route of the target. I did a bit of research into this. Very interesting algorithms. I tested in python notebooks.
  • PID (TODO). The PID determines what input to send to the drone control surfaces to hit the target (enemy drone, vehicle, etc.) at the predicted future location.

being an absolute newbie in both of the last 2 bullets, i'd need a team for support and lots of free time. I had neither (need a paying job). in any case, the key to stay at the forefront of the development is the control sw algorithms. the fiber optic spool and basic electronics is something the chinese will take over quickly (you cant compete with them; I worked there for years, speak and read the language quite well; what they are not good at is full systems engineering).


















24.0806

Great blog article.. talks about the future phase of the war. And using AI for long-term warfare.

image

FPV-Дроны с Машинным Зрением: Будущее Обороны

https://www.youtube.com/watch?v=1KKlGFHGG0U

Кофе с Любарским

https://www.google.com/search?q=prozorro+%D0%B1%D0%B5%D0%B7%D0%BF%D1%96%D0%BB%D0%BE%D1%82%D0%BD%D1%96+%D0%B0%D0%B2%D1%96%D0%B0%D1%86%D1%96%D0%B9%D0%BD%D1%96+%D0%BA%D0%BE%D0%BC%D0%BF%D0%BB%D0%B5%D0%BA%D1%81%D1%96&oq=prozorro+%D0%B1%D0%B5%D0%B7%D0%BF%D1%96%D0%BB%D0%BE%D1%82%D0%BD%D1%96+%D0%B0%D0%B2%D1%96%D0%B0%D1%86%D1%96%D0%B9%D0%BD%D1%96+%D0%BA%D0%BE%D0%BC%D0%BF%D0%BB%D0%B5%D0%BA%D1%81%D1%96&gs_lcrp=EgZjaHJvbWUyBggAEEUYOTIKCAEQABiABBiiBDIKCAIQABiABBiiBNIBCTYwMDE2ajBqN6gCCLACAQ&sourceid=chrome&ie=UTF-8

https://prozorro.gov.ua/framework/UA-F-2024-07-29-000001

image

image

⚠️ **GitHub.com Fallback** ⚠️