Discretizing Continuous BURLAP Domains?

[nharp...@gmail.com]

Hi! 

My name is Noah Harper and I am working through a school MDP/RL project and I'd like to use BURLAP's sample domains as examples. 

I've been tasked with selecting two MDPs and running Value/Policy iteration and one RL algorithm on them. I then have been asked to compare each algorithms performance. 

I would really like to try my hand at analyzing a more complex domain with 'many' states and so I was hoping to use one of the continuous domains implemented in BURLAP. The problem is, the value and policy iteration algorithms don't necessarily work over continuous domains. I know there is a function for discretizing continuous domains in BURLAP, but I'm having trouble using it, specifically with the Lunar Lander domain. 

If anyone is able to provide any insight into how I might discretize the Lunar Lander problem, it would be much appreciated!

Thanks, 

Noah Harper