Comment by fromthestart

I wonder if this is a tractable reinforcement learning problem. The objective is somewhat clearly defined-minimize spatial deviation over time from the input path.