After arriving in Belgium, I’ve started work on my PhD again. I cannot evaluate the latest experiments yet, as I have yet to have a computer for my own here, so I can’t give Waikato the information they need to allow me to SSH in.
So far, I have been finalising the decoupling process of the algorithm from PacMan. As far as I’m aware, it is complete, but only by testing the code on another domain will it prove to be so.
Kurt also gave me a Master’s Thesis on using RL to create a shepherding dog. I can use this as an example for seeing which way my research will take me. I’m figuring the relational path, but so far the PacMan experiment has been feature-based. Although, that could be converted into relational, which is something I should attempt next perhaps.