GuideNOW is an AI-powered 3D real-time robot vision guidance platform developed by Inbolt (France). It combines a 3D camera with proprietary Inbrain AI algorithms to enable industrial robots to dynamically locate workpieces and correct trajectories in real time — eliminating the need for mechanical jigs, fixtures, or precision indexing systems. The system is CAD-trained, requiring no physical part samples, and supports over 80 part references switchable within a single program.

How does real-time trajectory correction work?

The 3D camera, mounted on the robot arm in an eye-in-hand configuration, captures point cloud data at 30+ fps. The Inbrain AI engine processes this data, matches it against the stored CAD model or 3D scan of the target workpiece, and sends corrected coordinates to the robot controller — all within an 80 ms reaction loop. This is the fastest refresh rate in the industry for dynamic robot guidance. The system tolerates partial occlusion, varying lighting conditions, and works reliably in bright, dark, or uneven environments without additional illumination.

What makes CAD-trained AI different from traditional vision systems?

Traditional vision systems require physical part samples for template training, which can take hours and requires parts to be available. GuideNOW trains directly from CAD models or 3D scans in under 15 minutes — no physical samples needed. The Inbrain AI algorithm is reported to be approximately 100× faster than leading conventional 3D matching algorithms. Programming is done through GuideNOW Studio, a web-based interface requiring no computer vision expertise: mount the camera, import the CAD model, and configure the workspace — all achievable in under 30 minutes.

Which robots and applications are supported?

GuideNOW is natively integrated with Universal Robots (e-series, UR20), FANUC, ABB, and KUKA robots, with brand-specific execution licenses available. It is compatible with any industrial robot via standard Ethernet interfaces. The Intel RealSense camera is standard for the Real-Time variant; the High-Precision variant uses the Zivid 2+ M60 for sub-0.1 mm repeatability. Applications span pick-and-place, assembly (screw insertion, gluing), material handling, non-destructive testing, and semi-structured bin picking. The system handles both static workstations and moving assembly lines at up to 300 mm/s without encoders or presence sensors.

Frequently Asked Questions

What is the difference between GuideNOW Real-Time and GuideNOW High-Precision?

The Real-Time variant prioritizes speed with an 80 ms reaction time, 300 mm/s dynamic tracking, and 1 mm repeatability using an Intel RealSense camera. The High-Precision variant uses a Zivid 2+ M60 industrial 3D camera to achieve sub-0.1 mm repeatability for static high-accuracy applications, trading off dynamic tracking speed for precision. Part size range extends from 2×2 cm to 3×3 m across both variants.

Can GuideNOW work with moving assembly lines?

Yes. The Real-Time variant tracks workpieces on moving lines at up to 300 mm/s without requiring encoders or presence sensors. The system continuously updates the robot trajectory based on real-time 3D data, enabling robotic operations such as gluing, tightening, and assembly on continuously moving conveyors.