Sebi/Sim-Tello-KI-fun
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

add KI

Browse Source
This commit is contained in:
Sebastian Unterschütz
2026-04-21 11:35:04 +02:00
parent 8c77744cad
commit f5d344cf75
3 changed files with 48 additions and 91 deletions
Download Patch File Download Diff File Expand all files Collapse all files

121
README.md
View File

@@ -1,89 +1,46 @@
# Tello Drone Sim
# Tello AI Pilot (Extreme Performance Edition)
This is a simple simulation of a Tello drone using Ursina. The drone can be controlled via tcp calls.
Dieses Projekt ist eine hochoptimierte KI-Steuerung für die DJI Tello Drohne (und den integrierten Simulator). Es nutzt modernste Computer-Vision-Modelle via ONNX Runtime, um Personen zu verfolgen, Hindernissen auszuweichen und Ziele intelligent wiederzuerkennen.
In the repo there is the simulation server along with a client class that can be used to interact with the sim server
## 🚀 Hauptmerkmale
- **Hybrid AI Engine**: Nutzt **YuNet** (Face Detection), **MiDaS** (Depth Estimation) und **MobileNetV3** (ReID) – alle via **ONNX Runtime** für maximale FPS.
- **Multithreading**: Kamera-Stream und KI-Verarbeitung laufen in getrennten Threads. Das Videobild bleibt flüssig, egal wie schwer die KI arbeitet.
- **Visual Fingerprinting (ReID)**: Die Drohne erstellt einen digitalen Fingerabdruck einer fixierten Person und findet sie automatisch wieder, wenn sie kurz aus dem Bild verschwindet.
- **Intelligente Verfolgung**: Aktive Suche in Verschwindungsrichtung und aggressiver Verfolgungsmodus.
- **Echtzeit-HUD**: Professionelles Display mit Telemetrie, AI-Tiefenkarte und Radar-Zonen zur Hindernisvermeidung.
- **Rate Limiting**: Kontrollierter Datenfluss (10Hz) zum Tello-SDK für maximale Verbindungsstabilität.
![Features](./images/Features.gif)
## 🕹 Steuerung
Klicke mit der **Maus** direkt in das Videobild, um eine Person zu fixieren (**Lock-ON**).
## Setup
### Tastatur-Belegung:
| Taste | Aktion |
| :--- | :--- |
| **T** | **Takeoff** (Starten) - Einmal drücken und kurz warten |
| **L** | **Land** (Landen) |
| **M** | Wechsel zwischen **Manuell** und **KI-Modus** |
| **K** | Lock-ON Trigger aktivieren / Fixierung löschen |
| **Space**| **Not-Aus** (Stoppt alle Motoren sofort) |
| **W/S** | Vorwärts / Rückwärts (Manuell) |
| **A/D** | Links / Rechts (Manuell) |
| **R/F** | Steigen / Sinken (Manuell) |
| **E/Z** | Drehen Links / Rechts (Manuell) |
| **1** | Automatisches Drehen (Scan-Modus) an/aus |
| **Enter**| Programm beenden |
### Option 1: Dev Container (Recommended)
## 🏗 Architektur
- `drone_pilot/main.py`: Das Herzstück. Verwaltet Threads, UI-Events und die Hauptschleife.
- `drone_pilot/vision.py`: Die KI-Engine. Lädt ONNX-Modelle und verarbeitet Bilddaten.
- `drone_pilot/flight.py`: Der Flug-Controller. Berechnet RC-Vektoren basierend auf KI-Ergebnissen.
- `drone_pilot/ui.py`: HUD-Renderer für das OpenCV-Fenster.
- `drone_pilot/config.py`: Zentrale Konfiguration für Geschwindigkeiten und Schwellenwerte.
The easiest way to get started is using the provided dev container which includes all dependencies and GUI support:
## 🛠 Installation & Start
1. Installiere die Abhängigkeiten: `pip install -r requirements.txt`
2. Stelle sicher, dass die Modelle im Ordner `models/` liegen.
3. Starte das Programm:
- Simulator: `python run.py` (Default: `use_real_tello=False`)
- Reale Drohne: Ändere in `run.py` zu `use_real_tello=True`
1. **Setup the dev container for your platform:**
```bash
.devcontainer/setup.sh
```
This will auto-detect your platform (macOS, Linux, Windows, WSL) and generate the appropriate `devcontainer.json`.
2. **Open in VS Code:**
- Install the "Dev Containers" extension
- Open Command Palette (Cmd/Ctrl + Shift + P)
- Run "Dev Containers: Reopen in Container"
3. **Platform-specific requirements:**
- **macOS**: Install XQuartz (`brew install --cask xquartz`) and run `xhost +localhost`
- **Linux**: X11 forwarding should work out of the box
- **Windows**: Access GUI via VNC at `http://localhost:5901` (password: `vncpass`)
### Option 2: Manual Setup
If you prefer to set up the environment manually:
1. Create the virtual environment by running:
```bash
python3.12 -m venv venv
```
2. Activate the virtual environment by running:
```bash
source venv/bin/activate
```
3. Install the required packages by running:
```bash
pip install --trusted-host pypi.org --trusted-host files.pythonhosted.org -r requirements.txt
```
4. Export the python path by running:
```bash
export PYTHONPATH=$PWD
```
## Running the simulation
To run the simulation, run the following command:
```bash
python tello_sim/run_sim.py
```
You can try running some of the [examples](./examples) to see how the simulation works. The examples are located in the `examples` folder.
Or use the [client](./tello_sim_client.py) class to interact with the simulation server. The client class is located in the `tello_sim` folder.
## Troubleshooting
- For a specific python version on macOS, consider using [pyenv](https://github.com/pyenv/pyenv?tab=readme-ov-file#b-set-up-your-shell-environment-for-pyenv) to manage multiple python versions.
- Another alternative for macOS users is to use [Homebrew](https://brew.sh/) to install the desired python version:
```bash
brew install python@3.12
```
- Conda users can create an environment with the desired python version:
```bash
conda create -n tello-sim python=3.12
conda activate tello-sim
pip install --trusted-host pypi.org --trusted-host files.pythonhosted.org -r requirements.txt
export PYTHONPATH=$PWD
```
---
*Hinweis: Beim Fliegen mit der realen Drohne immer auf einen Akkustand > 15% achten, da der Startbefehl sonst vom SDK abgelehnt wird.*

2
drone_pilot/config.py
View File

@@ -8,7 +8,7 @@ class Config:
TARGET_PERSON_SIZE = 400
ALT_THRESHOLD = 0.12
YAW_GAIN = 0.12 # Reduced for smoother rotation
YAW_GAIN = 0.08 # Reduced for smoother rotation
FORWARD_GAIN = 1.5 # Kept high for fast pursuit
ALT_GAIN = 40

16
drone_pilot/flight.py
View File

@@ -69,9 +69,9 @@ class FlightController:
# Remember which side it was on
self.last_target_side = 1 if err_x > 0 else -1
# Rotation (Yaw) - FULL SPEED CAPABLE
# Rotation (Yaw) - SMOOTHER
if abs(err_x) > Config.FACE_DEADZONE:
yv = int(np.clip(Config.YAW_GAIN * err_x, -100, 100))
yv = int(np.clip(Config.YAW_GAIN * err_x, -50, 50))
# Forward/Backward pursuit - EXTREME SPEED
alignment_factor = max(0.4, 1.0 - (abs(err_x) / Config.FACE_ROT_ONLY))
@@ -85,18 +85,18 @@ class FlightController:
elapsed = time.time() - self.lost_time
if elapsed < 10.0: # Search longer and faster
yv = 80 * self.last_target_side
self.status = f"LOST TARGET: RAPID SCAN {'RIGHT' if self.last_target_side > 0 else 'LEFT'}"
yv = 40 * self.last_target_side
self.status = f"LOST TARGET: SCANNING {'RIGHT' if self.last_target_side > 0 else 'LEFT'}"
else:
self.status = "TARGET LOST: AGGRESSIVE PATROL"
yv = 60
yv = 30
elif self.smooth_face is not None:
# Face found but not locked
(x, y, w, h) = self.smooth_face
err_x = (x + w // 2) - (Config.WIDTH // 2)
if abs(err_x) > Config.FACE_DEADZONE:
yv = int(np.clip(Config.YAW_GAIN * err_x, -80, 80))
yv = int(np.clip(Config.YAW_GAIN * err_x, -40, 40))
self.status = "AWAITING LOCK"
else:
# Patrol mode - faster
@@ -105,8 +105,8 @@ class FlightController:
self.status = "PATROL: DASH"
fb = 40
else:
self.status = "PATROL: FAST SPIN"
yv = 60
self.status = "PATROL: SCAN"
yv = 30
return self._smooth(lr, fb, ud, yv)