RMS quality of a fitted space curve

Hello everybody,

I've got a point cloud with about 20 points describing a curve in a three-dimensional cartesian coordinate system. Now I need a fit which is represented by a list of curvature-torsion pairs over the length of the curve s (no spline - it's done by consecutively calculating the Frenet formulas) What I get from that is a new point cloud with about 1000 points I'm looking for a way to get the quality of the fit, so I can give it to a genetic algorithm as fitness function.

I was looking for a library but haven't found anything - neither on fitting frenet curves, nor on comparing two point clouds. Can anyone please help me?

Thank you very much in advance! Peter