This is a remix to show that the AI is highly unlikely to go back on itself, even if that is necessary to achieve the end goal, proving that AI that doesn't try all options is inherently imperfect. If you have any ideas comment below. However, if you could measure necessary distance to endpoint and use that, the AI would be able to progress to the end,
Thanks to the original project https://scratch.mit.edu/projects/259157608
