Extracting kernel density estimates to a new sample points in R

Question

I have spatial referenced data (shapefiles) of 400 polygons and points that are spread over them. I wish to create a raster of kernel density estimate for each polygon based on