Replanning Paths in a Dynamic Grid Environment

Finding a path efficiently is a challenging task in a commercial game environment. A typical game world will include numerous mobile entities navigating within the game world. A recently computed path for one mobile agent may need to be recomputed if the world changes or if another mobile obstacle has moved and obstructed the computed path. An obvious solution is to re-compute the path, based on information gathered from the previous search, thereby reducing the aggregate amount of effort spent on search. This is very likely more efficient than the current standard, which is to recalculate the path from scratch. The goal of this project is to determine whether replanning is worthwhile and practical in the commercial games domain. Problem Formulation / Problem Challenges In this project, we are primarily concerned with efficient path planning in a grid-like environment often seen in commercial games under typical system constraints. Pathfinding problems are ubiquitous in computer games. A game where nothing moves is, in general, very boring. A typical pathfinding request in a game would begin something like this: 1) The user selects a unit in the grid world 2) The user selects the desired location of the unit 3) The computer calculates a path from the unit's initial position to the desired position This problem is actually trivial, if not for the following constraints: 1) The path-finding request must be done very quickly (in milliseconds) The user should feel that the path-finding request is done promptly; this is especially true in games where response time is of paramount importance (RTS or FPS). 2) Limited system resources in terms of CPU and memory Game companies know that the lower the system requirements, the greater the market is for their game, because more people can run their product. Quite often, the game must be constructed for the lowest denominator. While a hard core PC gamer would regularly upgrade their computer to play the latest game, a console gamer does not have this option, as consoles are not upgradeable. The most popular gaming console now, the PlayStation 2, runs on a 300mhz chip. 3) Other mobile obstacles in the pathing environment In real-time strategy games (RTS) like WarCraft or Red Alert, there are numerous units that move in the gaming environment. Any of these mobile obstacles may disrupt any prior plans. Furthermore, the idea of "fog of war" introduces uncertainty in the pathing environment. It is worth nothing that the true aim of any pathfinding algorithm in commercial games is not necessarily to find the optimal path, but to not disappoint the user with poor A.I.