To interpolate linearly between the points in the table you should change your code to
rSetPos[1]:= arPos_1[index] + subindex*(arPos_1[index+1]-arPos_1[index])/20.0;
This will change your command curve from this:
to this:
This will improve matters, but in all likelihood the system performance will still be poor. One approach might be to break up your list of points into subgroups of 1024 and use FlexProfile. In this case the "motion law" for each section of the path would be based on a Point Table. Because you are dealing with large data sets, you will have to define your trajectory "on-the-fly" - begin by defining a FlexProfile based on the first 1024 points in your table. While this is running define a second FlexProfile based on the second set of 1024 points in your table and so on.
Another approach that we have successfully implemented in the past is to interpolate using cubic splines. In this case the command curve would be defined as a set of cubic polynomials. For example:
Here the coefficients are chosen so that the velocity curve is everywhere differentiable. The math here is not exactly trivial, but there are good resources available. A Practical Guide to Splines by Carl de Boor is excellent.
