Production ReadyAI-PoweredRobot Ready

See. Think.
Pick.

NKE Vision is an industrial-grade AI vision system that turns any robot arm into a smart picking machine, with real-time object detection, depth sensing, and millimetre-accurate coordinate output.

Capabilities

Everything Your Cell Needs

Purpose-built for industrial bin-picking and pick-and-place automation. Every feature exists because a production line demanded it.

🎯

Real-Time Object Detection

YOLO-powered inference at 20ms. Detects, segments, and ranks objects by size and centrality. No pre-labelling of your specific parts required.

📐

Millimetre-Accurate Depth

Stereo depth sensing fused with AI segmentation. Outputs X, Y, Z coordinates in robot frame. Ready for gripper path planning.

🖥️

Operator Dashboard

Clean, dark-mode web interface with live video feed, one-click trigger, system diagnostics, and coordinate readout. Works on any tablet or PC.

Simple Trigger Interface

The robot sends a trigger signal and instantly receives coordinates in response. No ROS knowledge required.

🔧

Alignment Grid Mode

Built-in crosshair overlay for on-site calibration. Set zero points, verify axis alignment, and confirm camera-to-robot frame mapping before going live.

🛡️

Diagnostic Matrix

Standardised error codes (E100–E302) broadcast to the UI and logged to console. Know immediately if it's a camera fault, model crash, or network drop.

Operator Experience

The Interface That Doesn't Get in the Way

Designed for operators wearing gloves, standing on a noisy factory floor, who need to know one thing: where do I pick next?

NKE Vision Operator Panel in action
Live annotated video with target overlay
One-click trigger with busy-state feedback
Real-time coordinate readout
System health at a glance
Event log with millisecond timestamps
Workflow

From Trigger to Gripper in 4 Steps

Seamless integration into any existing robot cell. No middleware, no complex orchestration.

1

Trigger

The system awaits a trigger from a controller, ROS topic, or UI button press, then begins detection.

2

Detect

AI model processes the current frame, segments all visible objects, and scores them by size and proximity to center.

3

Compute

Depth patch analysis around the target centroid. Median filtering rejects outliers. Pinhole camera model converts pixels to millimetres.

4

Deliver

Coordinates are returned as an X@Y@Z string. The robot parses the values, plans the path, and executes the pick. All within 500 ms.

Performance

Numbers That Matter on the Floor

No marketing fluff. These are the metrics that determine whether your cell hits cycle time.

< 20
ms Inference
YOLO + depth pipeline
±0.2%
Depth Accuracy
At 1–2 m working distance
24/7
Uptime Design
Auto-reconnect, watchdog

Ready to innovate your production line?

NKE Vision is deployable on your existing robot cell in days, not months. We handle integration, calibration, and on-site training.

Contact Us
NK Engineering logo in background