Adjustible Bayesian solver versus traditional A*
It uses maze precepts in the FEP algo to help pathfind but it does a similar iterative search to A*, with mileage varying, better for open mazes. The implementation needs some work but it does seem authentic.
Inspired by: https://eoinmurray.info/maze-solving