Run on-device YOLO object detection, ARKit 3D distance estimation, and local REST service coordinates in real-time. Purpose-built for robotic arm pick-and-place automation.
Powerful on-device processing designed to replace expensive industrial cameras with a simple iPhone.
Runs YOLO26x custom models locally. Optimized to recognize colored workpieces (red, green, blue, white) with adjustable confidence limits.
Combines 2D bounding boxes with ARKit spatial depth. Projects virtual rays to extract real-world coordinates and metric dimensions.
Runs a lightweight HTTP server directly on iOS. Exposes endpoints like /detect and /capture for immediate backend integration.
Configured to run on Apple Neural Engine (ANE) to maintain 60 FPS UI rendering, low GPU stress, and low thermal drift.
A robust coordinate transition pipeline from pixel coordinate space to real-world 3D metric coordinates.
YOLO identifies object boundary boxes on the screen to calculate their 2D pixel centers.
A ray is projected from the camera lens to the physical surface to determine depth coordinates.
Computes the actual object distance (m) and dimensions (cm) using hardware-accelerated vectors.
The FastAPI backend fetches the coordinate array over a local Wi-Fi Hotspot REST request.
Downstream applications (such as a sorting conveyor, robotic arm controller, AR visualization hud, or inventory database) consume coordinates instantly.
Watch the RobotVision app guide the 3DoF robotic arm through autonomous bin picking and placing.
Scan the QR code or click the download button to install the iOS app and start automating your workflow.
