Inverse Kinematics On Animation

Okay, so I've seen the existing locomotion system, but since I'm more interested in learning a way to make one of my own, I've come here to present my idea and get a little feed back and ask a few questions.

Right, so my current approach as it stands in my head (and on paper) is to have a script attached to the "heel" and "toe" bones of a foot that continually trace ray(this might be a little expensive, but I'll make sure to have it only collide with the "terrain" layer) and get the normal of the surface below it. If the "heel" or "toe" ever go lower than the "last" set normal, then the inverse of the intersection amount will be applied to the "heel" and "toe" bones and the average of the two intersections will be applied to the knee along with turning the normal into a rotation that will be applied to the "heel" and "toe".

In my head this seemingly plays out correctly, but I'm not sure if it will in reality. So , if there are any complications that you can think of off the top of your head that could ruin this current setup, please present them or if you have a better idea all together that could seemingly work I'd gladly appreciate the input.