Update readme with a demo usage of this model
Browse files
README.md
CHANGED
@@ -11,4 +11,32 @@ tags:
|
|
11 |
TinyLink is a **lightweight** language model fine-tuned to translate natural language instructions into MAVLink command dictionaries.
|
12 |
It is designed for edge robotics. Unlike solutions relying on cloud APIs, TinyLink runs **fully offline** on your local machine.
|
13 |
|
14 |
-

|
15 |
+
|
16 |
+
## Features
|
17 |
+
- Translates plain text instructions into MAVLink commands.
|
18 |
+
- Runs entirely on-device for enhanced privacy. No API keys or cloud dependency.
|
19 |
+
- Runs on everyday hardware; no GPU or excessive RAM needed.
|
20 |
+
- Tested with ArduPilot SITL.
|
21 |
+
- Achieves **1.5–2.2s** inference times on CPU, depending on hardware.
|
22 |
+
- Supported Commands:
|
23 |
+
1. Arm
|
24 |
+
2. Disarm
|
25 |
+
3. Takeoff
|
26 |
+
4. Land
|
27 |
+
5. Change mode (limited modes supported)
|
28 |
+
6. Move in X, Y, Z (Copter and Rover)
|
29 |
+
|
30 |
+
## Performance & Tested Platforms
|
31 |
+
|
32 |
+
| Platform | RAM | Inference Time (avg) | Status |
|
33 |
+
|-------------------------|------------------|-----------------------|-------------|
|
34 |
+
| Win 11 (App) & WSL2 (SITL) | 16 GB | 1.7 - 4s (Avg 2.2s) | ✅ Tested |
|
35 |
+
| Raspberry Pi 5 | 4 GB | 0.8 - 2s (Avg 1.5s) | ✅ Tested |
|
36 |
+
| NVIDIA Jetson Nano | - | - | ❌ Not tested |
|
37 |
+
|
38 |
+
|
39 |
+
## Demo
|
40 |
+
|
41 |
+
For a demo on how to use this model, you can check the following [Github repo](https://github.com/grep265/tinylink)
|
42 |
+
|