UCF CAP 5937 - nuWar - A prototype sketch-based strategy game

Unformatted text preview:

nuWar: A prototype sketch-based strategy game Greg Dunham, Ken Forbus, Jeffrey Usher Qualitative Reasoning Group Northwestern University 1890 Maple Avenue Evanston, IL 60201 USA {gdunham, forbus, usher}@northwestern.edu Abstract Today’s military strategy games provide unrealistic interfaces for players to interact with their units: Commanders don’t use mice and menus, they sketch. Developing strategy games currently involves grafting AI capabilities on top of a separate simulation engine, with new hand-crafted strategies for each game. We are experimenting with a novel approach for solving both problems. We started with nuSketch Battlespace, a knowledge-rich sketch understanding system developed for military users, and built a game engine, nuWar, on top of it. nuWar is a prototype two-player tactical war game which can be played either hot-seat or over a network. nuWar uses sketching as the primary way for players to express their intent to their subordinate commanders. The underlying ontology used by nuSketch Battlespace is used in both the simulation engine and in the bots which serve as subordinate commanders. We describe the architecture of nuWar, focusing on how it uses sketching, how the simulation engine is built upon the rich representational facilities of nuSketch Battlespace, and how the bots work. We discuss the tradeoffs we have found in this approach so far, and describe our plans for future work. Introduction Spatial reasoning is an essential tool for players of today’s strategy games. To succeed, a good player must be able to think about the game world more abstractly than as a simple set of tiles. A human player can carve the space up into a variety of qualitative spatial features, devising a plan of action with respect to these features [4]. Today’s games provide players with unrealistic interfaces for communicating such a plan to their units. Keypads, mice, and menus are often poor substitutes for a sketch when describing a spatial plan. In fact, military commanders have been doing this for years, sketching out battlefield plans, also known as courses of action (COAs) [5]. Yet it would be difficult or impossible for a general to Copyright © 2005, American Association for Artificial Intelligence (www.aaai.org). All rights reserved. communicate a spatially rich COA using the interface provided in most strategy games. The lack of qualitative spatial representations in strategy games also hinders the creation of game AIs [4]. Currently, AI infrastructure is typically built on top of a separate game engine, with strategies tailored to the engine’s internal workings [10]. After months of hard work, developers may be left with an AI architecture that is only useful for that one game engine implementation. The entire process must begin anew for the next game. This paper describes nuWar, a prototype two-player tactical war game that attempts to solve these two problems. nuWar uses sketching as the primary means of user interaction with the game. It is built on top of nuSketch Battlespace [5], a system for creating COAs, which has a rich underlying ontology. After introducing nuWar, we illustrate the sketching interface that we believe makes player interaction more natural than traditional war game interfaces. We then describe the advantages and Figure 1: The Blue-side player sketches orders for a direct frontal assault on the city of Dullsville.tradeoffs of the game’s deep ontological grounding. Next we explain our use of SADL, a declarative action language, to control the actions of subordinate commander AI bots. Finally, we discuss the current state of nuWar and our plans for future work. nuWar Overview nuWar is a head-to-head (Red vs. Blue) war game, which can be played either as a network game or in “hotseat” mode. The interface mechanic is similar to that used by commercial games such as Combat Mission. There is an initial scenario, either drawn from a library of scenarios or sketched by one of the participants. Each player simultaneously looks over the situation from their side’s point of view, and formulates a plan using the nuSketch Battlespace sketching interface. For example, the Blue-side player, whose sketch is seen in Figure 1, may see the pass between the mountains as the most direct path to seize the city of Dullsville. He sketches out his plan for a frontal assault on the seemingly unguarded city. The Red-side player, whose sketch is seen in Figure 2, cannot yet see any Blue units, but suspects that the enemy may try to seize the city by approaching through the pass. He sketches his orders for an ambush, laying in wait from well-concealed positions behind the mountains. Their plans complete, each player clicks a button indicating they are ready to execute the turn. When both players are ready, the simulator engine runs for a predetermined amount of game-time (Figure 3), and both players get to see what happened in the form of an animated movie that uses the same graphical elements from the battlespace sketch. Figure 4 shows the end-state of the turn-movie generated from the execution of the turns described above. The underlying physics model and task execution details used to generate the turn results are described below. After the movie has been viewed, players can modify their plan, if desired, and initiate another round of game simulation. The motivation for developing nuWar was to provide a game-based environment for knowledge capture from military experts. That is, by “watching” two experienced officers play each other in the game, we plan on building up a library of strategies and tactics. This means that the constraints on realism in the current version are tighter than might otherwise be desirable for mass-market gameplay, e.g., the large number of units, kinds of units, and their capabilities. Sketching is a player’s sole means of affecting the state of a nuWar game. A key consideration while designing the sketching interface was balancing the tension that arises Sketch KB Simulation Engine Bot AI Physics EngineFigure 3: Architecture diagram illustrating the process used to generate each discrete "tick" of a turn movie. Figure 2: The Red-side player describes an ambush by sketching a course of action. Figure 4: The results of the turn illustrated in Figure 1 and Figure 2. The Red ambush succesfully prevented Blue from seizing the city of Dullsville. Shaded areas represent regions that


View Full Document

UCF CAP 5937 - nuWar - A prototype sketch-based strategy game

Documents in this Course
Load more
Download nuWar - A prototype sketch-based strategy game
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view nuWar - A prototype sketch-based strategy game and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view nuWar - A prototype sketch-based strategy game 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?