OK, here is a few bugs I've found.
1. Workbot can't find current beacon if another workbot is staying on it. (After start of algorithm left-bottom workbot will go to upper beacon)
2. Workbot can't find nearest water from lake (I don't know why, because another workbot can do this). From the other hand he can find water from sea.
3. And there are some artefacts on screen sometimes. I think it's when too many elements are on the screen at the same time.
You're issue 2 is probably due to the command "Find nearest Water ?" not being based upon a Beacon but instead being based on the location that the bot was when it was trained, as indicated by the ? symbol, while the bot is now too far from that training location and the training location being too far from any water. It could also be due to this bot being loaded with commands from data storage that had been saved from another bot whose own training location, as indicated by the ? symbol, is too far from this bot . You can retrain this bot and click on that ? symbol to use a nearby Beacon instead of the initial training location for finding the nearest water.
Another problem is that all of the Folk's huts are so close together that if the nearest water is on the opposite side of all of the huts from the bot that there is not sufficient time for the program to find the shortest pathway through all the huts to the water. When you place the huts, you should provide passage ways through them. Leave some gaps between the huts (and other buildings) so that bots can pass through them and the program can find a pathway.
I do not think that the lack of spaces between the huts will "prevent" all bots from walking to the water. I meant that where this bot is located relative to the water might be south of the huts. The red circle is showing the area that the program will search for the nearest water to find the shortest path, and that area includes about 100 huts. The program would calculate the lengths of the paths including those through the huts, but since the only paths through them are moving east-west between the rows (rather than moving north through non-existent gaps between huts), the amount of time to calculate all of those times for thousands of combinations of pairs of the 100 huts is so large that this bot doesn't move. If there were gaps, the program would find paths using them in a shorter amount of time and then could find the shortest path and thus the "nearest" water. The other bot that you mentioned no doubt has nearby water that is not reached by a path through the blockade of huts, and that long calculation and comparison of times would not be as great.
