In case you have any remarks or questions they are always welcome at either the education committee, via the slack channel ec-helpme, or the Board via education@serpentineai.nl.
In this lesson we are going to talk about the following things:
The GitHub links for this lesson are: Browse, Zip, Diff.
The links are currently only available for members.
As mentioned in the previous lesson, we are only taking into account that the end position of a path is safe. In order to make sure that the positions we pick are absolutely safe we will have to filter out all paths that aren't safe to walk on. For this we have to update find_reachable_safe_locations. Think about the changes and apply them.
A possible solution path is:
end=(-1, -1)).def find_reachable_safe_locations(self, board: np.ndarray, danger_map: np.ndarray, location: tuple) -> list:
came_from, came_from_direction = self.pathfinder.bfs(
board=board, start=location, end=(-1, -1), passable=self.passable)
safe_positions = [position for position in came_from if danger_map[position] != 1]
safe_positions.sort(key=lambda position: sum((pos - loc) ** 2 for pos, loc in zip(position, location)))
safe_paths = []
for safe_position in safe_positions:
path = self.pathfinder.reverse_path(came_from, came_from_direction, safe_position)
path_position = location
path_is_safe = True
for turn, direction in enumerate(path, start=1):
path_position = tuple(np.array(path_position) + direction.array)
if 0 < danger_map[path_position] <= turn:
path_is_safe = False
if path_is_safe:
safe_paths.append(safe_position)
if safe_paths:
return safe_paths
return safe_positions
Now that we have implemented all the primary logic of staying safe it is time to become more aggressive. It is time to pick up powerups and start to 'remove' the opposition.
As you might have noticed there is no coding solution for this part (or the following), it is now all up to you. By now you have more than enough knowledge of python and git to solve this and many more problems by yourself. If you do get stuck, feel free to contact the education committee via slack or email.
There are many more features that you can implement, see if you can create the ultimate pommerman bot. Some ideas for features are:
kick a bomb to an enemy).As you might have noticed there is no coding solution for this part (or the previous), it is now all up to you. By now you have more than enough knowledge of python and git to solve this and many more problems by yourself. If you do get stuck, feel free to contact the education committee via slack or email.
In this lesson we updated the pathfinding of the bot, to be as safe as possible. With all this safety it is up to you to further improve the bot, to be the most dominating bot. Also this lesson is very open to implement things yourself. As a final tip, always try to first implement something in a simple way, before making it more and more complex.
Thank you for sticking with us, up to this point. At the moment of writing there have been no further request to continue this tutorial series, and therefore it was decided to stop the series here. This is not the end, there are still a lot of ways to keep improving the bot and we encourage you to keep exploring the oppertunities.
If you enjoyed the series or have any feedback please tell the education committee about it, we always like to hear your view of the lessons. And maybe if there are enough requests or good ideas we might continue this series one day.
With all that said, the adventure is by far not over, check out our education page for more content, or contact the education committee for specific project requests or ideas.