AI-powered visual inspection that detects machining defects, cutting-tool fractures, and finished workpiece burrs before they become expensive scrap—engineered with zero CNC controller modification.
In high-precision CNC manufacturing, physical tool failure occurs within microseconds. Unnoticed micro-chipping on cutting inserts escalates kinetic friction, resulting in damaged workpieces, out-of-tolerance surface finishes, and devastating spindle rebuilds exceeding $15,000+.
Conventional vision setups require intrusive PLC re-wiring, complex fieldbus taps, or continuous cloud connectivity. Sightline Edge AI acts as a non-invasive optical sentry—monitoring cutting tools and output feedbeds with real-time mathematical vision and drop-in deep learning, functioning 100% air-gapped without touching CNC controller logic or voiding manufacturer warranties.
Core business logic is compiled into native C-machine shared objects (.so) via Cython. Substantially increases reverse-engineering resistance and eliminates raw Python scripts from host disks.
High-resolution 2K captures stream into /dev/shm shared memory. Coupled with atomic POSIX staging and reactive FIFO purging at 85% capacity, flash storage endurance is protected for continuous multi-year operation.
Zero cloud dependencies. Runs native mathematical optical telemetry out-of-the-box, with an autonomous drop-in provision for INT8 quantized ONNX neural models (PatchCore / YOLO) for customer-trained defect classes.
Asymmetric cryptographic licensing bound to hardware motherboard/CPU SHA-256 digests. Monotonic database RTC verification strictly prevents system clock rollbacks on air-gapped IPCs.
Simultaneous dual-ROI cropping: Zone A inspects spindle tool condition, chipping, and coolant spray; Zone B conducts Automated Optical Inspection (AOI) on deliverable parts for physical burrs.
Engineered with SQLite Write-Ahead Logging (WAL) and synchronous=NORMAL. Survives sudden factory floor electrical blackouts without database locking or metadata corruption.
Sightline is a software appliance. Compatible with standard factory imaging hardware (Cameras supplied separately):
| Optical Interface | Supported Protocols | Target Camera & Sensor Hardware |
|---|---|---|
| Industrial RTSP Stream | RTSP / H.264 / MJPEG over TCP | Hikrobot, Dahua, Basler IP Cameras, Uniview Industrial |
| USB3 Industrial UVC | DirectShow / Linux V4L2 High-Bandwidth | Basler dart/pulse, FLIR Blackfly, Imaging Source USB3 |
| Embedded Sensor Modules | MIPI-CSI / Sony IMX 2K Sensors | Raspberry Pi HQ Camera, Arducam IMX477, Jetson Nano CSI |
| Legacy Machine Retrofit | Direct USB 2.0 Optical Feed (1080p/2K) | Standard factory floor optical retrofits & articulated arm cams |
Sightline Edge Vision AI is protocol-agnostic and designed to interface with a broad spectrum of industrial, IP, and embedded camera sensors. Optimal defect detection depends on working distance, shop-floor illumination, and lens magnification rather than raw resolution alone.
Best suited for permanent machine shop enclosures. Supports RTSP/ONVIF streams over long ethernet runs without signal attenuation in high-EMI factory environments.
Engineered for close-range tool spindle monitoring. Direct UVC/V4L2 connection delivers uncompressed 2K frames with zero network latency for high-speed cutting.
Ideal for pilot evaluations, articulated robot arms, and space-constrained inspection trays utilizing Sony IMX high-sensitivity sensor modules.
Review our verified compatibility blueprints for lenses, industrial enclosures, and illumination hardware suitable for Sightline deployments.
*Notice: Some hardware recommendations contain verified partner links. Purchasing through these adds no additional cost to you and supports ongoing platform R&D.
Select your machine operating system below. Enter your work email in the respective tier to receive your dedicated standalone package link directly in your inbox.
Self-contained embedded engine for Windows 10/11 IoT & 64-bit HMIs. No Docker or Python required. Includes 1-click START_SIGHTLINE.bat.
Self-contained portable engine for fanless Ubuntu/Debian Industrial PCs (IPCs). No Docker daemon required. Includes 1-click ./start_sightline.sh.
.ZIP into any local directory.START_SIGHTLINE.bat to initialize the microservice.http://localhost:8000. If your browser displays "Unable to connect" for a brief moment, simply click "Try Again" or press F5 while the background engine warms up.tar -xzf cnc-sightline-v1.0-linux.tar.gz../start_sightline.sh.http://localhost:8000.license.key into the application folder to instantly activate.Permanent cryptographic license bound to your machine hardware. Zero recurring software fees. Software appliance only (Cameras supplied separately). Multi-node volume tiers available for 3, 5, or 10-machine plant suites.
START_SIGHTLINE.bat (Windows) or ./start_sightline.sh (Linux).
http://localhost:8000 in your browser and copy the unique 16-character Hardware Fingerprint displayed on screen.