The CNC moment for surface repair is here
Modern repair and remanufacturing facilities run high-mix, low-volume operations. The parts arriving on the shop floor: pipes, housings, brackets, curved structural panels, freeform castings; rarely look like the part that came in yesterday. Yesterday’s robots are not designed to manage this variation.
Traditional industrial automation was built for repeatable, fixtured production lines. In a repair environment, geometry varies from job to job, fixtures are improvised, and every new surface demands new programming. The result is a paradox familiar to any operations leader: a million-dollar robot quietly sitting idle while a skilled artisan grinds, sandblasts, or sprays by hand because reprogramming the work cell would take longer than just doing the job.
PickNik Robotics has spent the last several years closing that gap. Our latest work pushes robotic surface treatment past a key inflection point, one we believe is analogous to what CNC did for machining. Instead of teaching a robot where to move, an operator points and clicks at a live camera view, and the system figures out the rest.
Scan-and-spray, in one workflow
PickNik’s scan-and-plan architecture, built on our commercially available MoveIt™ Pro software platform, turns a robotic arm into an adaptable surface treatment tool that’s easy to run.
The workflow is deliberately simple: an operator brings the robot into the work area, scans the part with a wrist-mounted depth sensor, selects the region of interest directly on the live camera feed, and reviews the auto-generated coverage path in a digital twin before pressing go. No waypoint programming. No CAD model required. No robotics PhD on standby.
What’s happening under the hood
- Perception-based surface reconstruction: a wrist-mounted depth sensor captures point clouds and registers them into a unified world frame, using iterative closest point (ICP) merging across multiple scan poses where needed. The resulting mesh is the geometric ground truth the system plans against.
- Curvature-aware coverage path planning: surface normals are computed directly from the reconstructed mesh, and a raster trajectory is generated that holds tool standoff and orientation constant across cylinders, concave panels, and freeform shapes — including point-and-click avoidance of features like bolt holes or fasteners.
- Constrained-environment motion planning: the same scan that builds the surface model also builds the collision world the robot needs to safely operate inside tight, partially-enclosed geometries. Reachability analysis and full digital-twin previews let the operator confirm the plan is collision-free before any motion is commanded.
- Human-in-the-loop UI: an intuitive interface abstracts robotics complexity for shop-floor operators while preserving power-user access to standoff, tip speed, step-over, layers, and pattern parameters for advanced users.
From simulation to working hardware
This isn’t a slideware concept. The full scan-to-execution pipeline runs end-to-end on collaborative robot hardware, with a working operator UI and validated coverage on representative shapes: flat surfaces with irregular regions of interest, cylindrical pipe sections, spherical surfaces, long concave panels, and irregular freeform parts.
Coverage paths planned in simulation were transferred onto physical hardware with no manual waypoint programming and no measurable surprise — the digital twin accurately predicted robot behavior prior to every hardware run, which is exactly what you want when the next step is deploying on a customer’s shop floor.
Equally important, the operator interface was reviewed with frontline maintenance personnel, not just engineers. The point-and-click coverage selection, screw-hole avoidance, and operator UI are direct responses to what people doing the work actually ask for.
One platform, many surface treatments
The team built this for robotic cold spray repair — a solid-state metal deposition technique that restores corroded or damaged components without the thermal distortion of welding, but the architecture was designed to be multi-purpose. The same scan-and-plan core handles any process that needs to follow a surface with a tool:
- Cold spray and thermal spray repair
- Sandblasting and surface prep
- Painting and coating
- Grinding and finishing
- Welding (where path-following is the dominant motion problem)
Because the planning and control layers are robot-agnostic, the same software stack runs on different arms, in different cells, doing different jobs, without rebuilding the application from scratch each time.
Why this matters
The business case for flexible robotic surface treatment is straightforward, and it shows up in four places at once:
- Reduced setup time. New parts and new repair scenarios don’t require a programming campaign. Setup collapses from hours to minutes.
- Higher throughput. Robots that previously could only handle simple flat-raster jobs are now usable on curved and irregular geometries, expanding the share of work that can be automated.
- Consistent quality. Standoff, orientation, and tip speed are held to spec across every job, removing the variability that comes with handheld operations.
- Ergonomic relief. Sandblasting, grinding, and handheld spray operations are physically demanding and a leading source of repetitive-strain injuries. Automating them protects skilled workers and lets them focus on judgment-driven tasks.
Put together, this is what flexible automation has been promising for two decades: capital equipment that adapts to the work in front of it, not the other way around.
Where this is going
The same scan-and-plan core that drives surface repair today is the foundation for a much broader class of robotic applications — including dynamic inspection and on-the-fly remanufacturing, where the environment can’t be locked down and the work can’t be pre-programmed. We see this as a foundational shift in how applied robotics gets built: operators specify intent, the system handles motion.
PickNik builds MoveIt Pro, the production-grade development platform and runtime engine behind this work. It integrates the latest ML models with classical robot control to deliver advanced robotic applications quickly and with the guardrails that real production environments require.
MoveIt Pro is PickNik Robotics’ commercial robotics software platform that extends the capabilities of the open-source MoveIt project. Designed for unstructured environments, MoveIt Pro provides powerful tools for motion planning, grasping, perception, and behavior sequencing. With pre-built robot behaviors and a modular runtime architecture, MoveIt Pro significantly reduces engineering effort and time-to-deployment. Its hardware-agnostic design supports a wide range of commercial and custom robotic arms, making it ideal for scalable automation across space, manufacturing, and logistics sectors.
