VISION 2026: 3D Vision and Live Motion Planning

VISION is the machine vision trade fair to go to. Happening every two years in Stuttgart, it draws the people who build robotic cells. For this edition, Zivid showcased its 3D cameras and a brand-new product. Watch the recap if you don't have time to read:

Zivid 3D camera portfolio covers all applications

We brought the full Zivid portfolio, the Zivid 2+ R-series and the Zivid 3, with live point clouds on screen. Our goal was simple: show visitors the best 3D and 2D data they could find at the fair, on the kinds of objects that usually cause trouble.

The R-series models cover different jobs: the MR60 targets assembly, robot guiding and inspection, the MR130 and LR110 piece picking and bin picking. The Zivid 3 XL250 is built for stationary mounting above large volumes such as pallets, truckloads, and deep factory bins.

Introducing Zivid Motion to the public

Zivid Motion is our motion planning software for robotic manipulation. It plans collision-free robot paths in real time, using 3D vision to understand the cell as it actually is rather than as it was taught.

At the booth, a UR robot with a Zivid 2+ MR130 camera and ArUco markers worked around a pink obstacle structure. Visitors were invited to pick up the obstacles and move them. Each time, the robot replanned and continued its touch tasks with submillimeter precision, without reteaching, extra code branches, or reprogramming. A screen next to the cell showed the planner's view of the scene in real time.

What the demo showed in practice:

• Paths in under 100 ms. Zivid Motion generates executable paths in under 100 ms, even in cluttered spaces.
• Full arm collision checking. It checks the whole arm, not only the tool, against a collision model built from CAD, meshes, and point clouds.
• Tighter paths. Continuous 3D data lets the robot plan close to obstacles instead of taking long detours, which helps cycle time and workspace use.
• Less commissioning work. Cell mapping, calibration, and validation are automated.

It was something visitors had not experienced before: a robot reacting to a changing cell without anyone touching the robot program.

Customer demos

Two partners showed how they build on Zivid cameras in their own systems.

Austrian Institute of Technology (AIT). AIT presented MARI, Model Based Adaptive Robotic Inspection. MARI is a flexible inspection framework that plans robotic inspection workflows from CAD models. In the demo, a robot scanned a metal housing from several viewpoints, merged the scans into one point cloud, and compared it against the reference model as a deviation map.

EVT Eye Vision Technology. EVT demonstrated 3D vision for pick and place guidance with its EyeVision software. An overhead 3D camera located miniature shipping containers on a table, and a robot arm picked and placed them using the 3D pose shown on screen.

In both cases, the partner focuses on its application software and relies on Zivid for the 3D and 2D data.

Why motion planning needs excellent 3D vision

Seeing difficult objects reliably was the first barrier in robotic picking. Motion is the next one. But a motion planner is only as good as the scene it plans in: if the point cloud misses a transparent bottle or a dark bin edge, the robot either collides or takes slow detours to stay safe. Running an advanced planner on poor 3D data is like putting a cheap motor inside a Ferrari. That is why we showed Zivid cameras and Zivid Motion together at VISION: excellent vision is what lets the planner take the short, safe path, every cycle.

Missed us in Stuttgart? Explore Zivid Motion or compare our cameras.