Micro-patterned surfaces reduce drag, repel water, and resist fouling. The patterns are tiny — micron-scale grooves or dimples that alter how fluids interact with the surface. Making them is straightforward with expensive precision machinery. Making them at scale — on ship hulls, building facades, industrial pipes — is not. A laser engraver that costs hundreds of thousands of dollars and raster-scans at millimeters per second cannot cover a surface measured in square meters in any practical timeframe.
The solution isn't a better machine. It's more machines, each cheap enough to deploy in numbers. Multiple robots, each carrying a patterning tool, coordinate to cover a surface. The coordination problem — how do the robots divide the work without gaps or overlaps? — is solved by ergodic control, an algorithm that specifies coverage objectives as probability distributions rather than fixed paths.
Each robot broadcasts a compressed representation of where it's been. The others adjust their trajectories to cover the areas that remain unvisited. No central planner assigns regions. The division of labor emerges from the algorithm's structure: each robot maximizes its own ergodic coverage while incorporating information about what the others have already done.
The result: robot-produced micro-patterns measurably lower the coefficient of friction on metallic surfaces. Not as precisely as the lab machine, but precisely enough. The quality gap between artisanal and distributed manufacturing is smaller than the scalability gap between them.
The through-claim is about when precision stops being the bottleneck. The lab technique produces better patterns but can't produce enough of them. The distributed technique produces adequate patterns everywhere. In manufacturing, coverage often matters more than perfection — and coverage is what parallelism provides.