Some of the problems addressed yesterday have now been fixed (theoretically). The JUnit tests say they work, but until I run a lab test, I won’t know for certain.
– I-piece rotation straight away. This has been fixed to make it IDLE first then rotate next step. This applies for every piece also. Just in case it’s more than I-piece.
– I-piece bad exploratory placement. This has been modified to place I-pieces next to climbs or drops of 2 or by the wall. Not perfect as it disregards climbs of 3 or more that aren’t special states.
– Find piece problem. May have been fixed by setting the prevObservation to null at agent_end. Unsure though. Lab test to check.
Need to play around with stuff at the lab. Having to go to the lab all the time is rather annoying, so I’m getting Linux here to run it on.