Is it possible to make a Reinforcement Learning agent for path planning in a 2D grid where visiting certain intermediate states is mandatory?
Please give an insight if it is indeed possible as to how I can proceed about it or give some reference.
I mean how to achieve it. I have tried giving more rewards to each of the intermediate states (grid locations in our case) that are mandatory to visit. But I want to visit certain states in a certain order. For example, Let's say S1 S4 and S5 are intermediate states that are mandatory to visit in a given order, S0 is the initial state and SG is the goal state. Then S0 -> SI -> S1 -> SI -> S4 -> SI -> S5 -> SI -> SG must be a sequence where SI = (zero or more intermediate states). The method mentioned above will make an agent visit given states but not necessarily in the order that I want.