i underestimated how tedious reya is

i am still in the first step of the project (not to say that it is minor, it arguably is 80% of the project), which is spatial memory. I had just began focusing my effort on it a few days ago, most of the last 2 weeks were focused on technical details of the reservoir computing backend.

right now the pipeline is supposed to look like:

1. generate a K x K random tiny cave via cellular automata

2. pick N random positions and angles

3. cast 100 rays in a 90 degree FOV on that grid

4. record which grid cells are being hit at each N

5. for each N, create a vector contianing the theta, position, distances, and a vector of length K * K in which a 1 represents a cell that was hit by the rays. the second vector is accumulative, such that all the cells hit by the previous samples (each N is refered to as a sample) presist.

this pipeline is very complicated, error-prone, and has been the source of my headaches ever since the project started.

however, this model of spatial memory it is not biologically plausible and the way it handles vision would break down once entities are introduced, since that would require actual vision. as such, i might scrap that idea and move on to a different model of spatial memory that incorporates actual vision and isnt incompatible with entities like enemies or props.