a classic greedy algorithm merging the breadth-first search approach with a heuristic evaluation function to prune nodes on each level of the tree… Beam Search is considered a good choice for games that have limited interaction with the oppo- nent(s), so the planning part is more important than the exact opponent prediction - wikipedia

At the cost of losing the optimality, the algorithm allows us to easily trade branching factor for increased search depth – thus gain an advantage on future planning.

References

• Developing a Successful Bomberman Agent
• Beam stack search
• (CG) Fall Challenge 2020 - Agade BFS pseudo code & Zobrist hash