Robotic Sanding at Sub-Millimeter Accuracy with Machine Vision

01 Story

Sanding, polishing, grinding and surface treatment are originally jobs of the hand. Across automotive, aerospace, marine, additive manufacturing, composites and furniture, manufacturers have long lived with the consequences: variable quality, high operating costs, capacity limited by skilled labor, and finishing lines that simply cannot scale with demand.

Supero was born from that gap. The platform is an advanced robotics software stack designed to automate industrial surface finishing on parts whose geometry refuses to sit still. Instead of asking operators to teach robots point by point, Supero asks the cell to look first to understand the part in front of it, then plan the work.

The vision is to make robotic finishing flexible, accessible and efficient: shorter setup, consistent quality, and far less reliance on manual labor. Crucially, Supero was built to work across sectors. Wherever surface quality is critical - metallic, composite or additively manufactured - the platform can adapt to the part rather than the other way around.

02 Challenges

Surface finishing is one of the hardest manufacturing operations to automate. It demands the judgment of a skilled hand, the consistency of a machine, and the flexibility to deal with parts that never arrive twice in the same orientation. For Supero's team, building a platform that could take this on meant confronting a stack of problems that have kept finishing lines manual for decades:

• Complex, variable geometries. Parts in automotive, aerospace, marine, additive manufacturing, composites and furniture rarely share the same shape twice. Traditional robotic programming, which assumes a known geometry, breaks down quickly.
• Imprecise part presentation. In real production environments, components cannot be jigged with the sub-millimeter precision robots typically require. The cell has to figure out where the part is, not the other way around.
• Sub-millimeter accuracy requirements. Path planning, collision avoidance and finishing quality all depend on geometric data that is accurate to fractions of a millimeter - anything less and the tool either misses the surface or hits it the wrong way.
• The need for color, not just shape. Finishing isn't purely geometric: identifying relevant areas, segmenting the part from background clutter, and spotting defects all require color information aligned with the 3D data.
• Industrial cycle times. Even perfect data is useless if it takes too long to acquire and process. Surface finishing automation has to keep pace with production lines, not slow them down.
• Safe motion in tight, occupied spaces. Robots have to plan trajectories that avoid singularities, self-collisions and unintended contact with the workpiece — while still reaching every point that needs to be finished.
• Inaccessible to non-experts. Traditional robotic programming requires specialist knowledge. For finishing to scale, shop-floor operators need to be able to set up and run a job without becoming robot programmers themselves.

Each of these challenges is solvable in isolation. Solving all of them together, reliably, at industrial cadence, is what makes robotic surface finishing genuinely hard — and what Supero set out to address.

03 Solution

Supero acts as the intelligence and orchestration layer of the robotic cell, tying together 3D acquisition, object reconstruction, path generation, motion planning, process execution and operator interaction in a single workflow. A typical Supero cell includes:

• Industrial or collaborative robot arm
• Surface finishing end-effector — sander, polisher or grinder
• Zivid 3D camera (Zivid 2+ LR110)
• Robot controller
• PLC and safety system
• Industrial PC running the Supero software platform
• Tooling and workholding systems
• Safety enclosure and interlock systems
• HMI / operator interface

The 3D camera choice was decisive. Running the Zivid 2+ LR110, the team singled out one foundational capability: a single coherent dataset of 3D coordinates, 2D images, color and surface normals, all pixel-aligned, removing a class of plumbing problems for algorithms built from scratch.

The LR110's 700–1700 mm working distance lets Supero handle parts of very different sizes and shapes with operators free to position objects loosely, and Zivid's built-in hole repair, noise and outlier filters deliver cleaner point clouds at acquisition time. As a result, 3D vision did not dominate the project timeline; the bulk of the effort went into robotics integration, safe motion planning and industrial workflow.

What sets Supero apart is the combination: advanced robotics, unified 3D and 2D vision data, intelligent path planning and operator-friendly tooling in one platform. Operators don't need prior knowledge of the part, and they don't need to interact with the robot directly. The platform also works both at the cell and remotely: engineers can prepare and simulate machining offline, and the machine-side operator simply loads the project and runs it.

04 Results

Because Supero targets surface finishing rather than pick-and-place, the indicators that matter are different. The platform has been deployed and validated across industrial finishing scenarios involving complex geometries and very different manufacturing sectors. In every case, the same themes show up:

• Robot programming time falls sharply, because the operator no longer programs the robot.
• The same cell handles different geometries without re-engineering the setup.
• 3D reconstruction and path planning are reliable enough to trust at production cadence.
• Manual finishing effort drops meaningfully, with consistency and repeatability rising in step.

End-customer sentiment matches the engineering picture. Industrial partners highlight the simplicity of use, the reduction of manual robot programming complexity, the flexibility on variable geometries, and the ability to automate processes that have traditionally been deeply manual. The offline preparation workflow - design and simulate in one place, execute in another - is consistently called out as a productivity unlock.

Supero was designed from day one as a scalable, internationally deployable platform, and the roadmap reflects that ambition. The long-term goal is a flexible intelligent robotics platform that can take on complex surface processing across an ever-widening range of industries and geometries with Zivid vision continuing to provide the foundation it sees the world through.