ociaw / roversim Goto Github PK

Core currently offers two different level generators - DefaultLevelGenerator and
MazeLevelGenerator. At the moment, these can only be switched at compile time by changing the
referenced class. This should be a runtime setting, using a dropdown box or something similar.

This AI will bring the path finding of the Mark II to the modern API. It'll cache path finding results to make simulation much quicker, and it will be more intelligent about where it decides to go.

Currently we use System.Random. However, while it does take a seed, Random isn't actually guaranteed to return the same numbers across different versions of the runtime. In addition, it has flaws that mean the numbers it generates aren't as high quality as intended. This makes a strong case for moving to a high quality, reproducible RNG instead.

Core should have unit tests to help ensure correct behavior. AIs themselves probably don't need tests however, and clients are generally just a wrapper around Core anyway.

I'd like to be able to view and share AI and simulations on the internet easily. Ideally, I'd use Blazor to serve the assemblies and do all the work on the browser so I could just host it on like any old static site. If necessary, however, an ASP.NET Core API could be created to do simulations and then render the results client side using ordinary JavaScript.

The Avalonia Client currently can render terrain correctly. However, it does not have the ability to simulate the rover and update the terrain as it progresses. This will happen once I can figure out how to properly do this using ReactiveUI/ReactiveX.