← All projects
PRJ / 10 · Edge AI motion control
Edge AI Long-Range Object Tracking Mount
A dual-mode pan-tilt mount combining quantized object detection, centroid-PID tracking, encoder feedback, and sidereal-rate compensation.
01 / Overview
A two-axis tracking platform supporting terrestrial object tracking and telescope astrophotography compensation.
Objectives
- Detect targets with an edge-optimized model
- Track detected centroids smoothly
- Support terrestrial and sidereal-rate modes
My role
[Add your exact role, ownership, dates, and collaborators.]
02 / Problem
Long-range visual tracking requires efficient on-device detection and smooth closed-loop positioning across terrestrial and astronomical modes.
03 / System architecture
01Camera input
02Quantized MobileNet SSD
03Centroid-PID loop
04Encoder-controlled pan-tilt mount
Hardware
- Two-axis pan-tilt mechanism
- Micro-stepped stepper drivers
- Encoder feedback
- [Add edge-compute platform and camera]
Firmware / software
- Quantized MobileNet SSD
- Centroid-PID tracking
- Terrestrial tracking mode
- Sidereal-rate compensation
04 / Development process
Process
[Add model, mechanical, calibration, and tracking-test stages.]
Challenges
[Add inference, target stability, positioning, and optical challenges.]
05 / Results
The prototype combines edge detection with closed-loop two-axis tracking in two operating modes. [Add verified accuracy and test conditions.]
06 / Image gallery
IMAGE / 01[Add prototype photograph]
IMAGE / 02[Add detail photograph]
DIAGRAM / 01[Add architecture visual]
Technologies used
Start a conversation
Have something meaningful to build?
Connect with meEmbedded products, robotics, artistic installations, technical visuals, or stargazing experiences—tell me what you're imagining.