Computing optimal ways of achieving these goals may involve optimization

Computing optimal ways of achieving these goals may

This preview shows page 76 - 78 out of 139 pages.

Computing optimal ways of achieving these goals may involve optimization algorithms at a low level (e.g., find the shortest path from here to there) progressing up to complex planning strategies at a high level (e.g., assemble a bunch of wooden crates in order to form a stable structure making it possible to climb out of a pit). Often in games, AI is most evident in games when it fails, that is, when nonplaying characters behave in an inexplicably nonsensical manner. (For example, a pedestrian character that continues to walk nonchalantly down the street in the midst of a gun fight.) Roles of Game AI: Generally, AI is used in games is to determine complex behaviors that not specified by the player nor a direct effect of physics. Examples include: Nonplayer Opponents: In a first-person shooter game, opponents should exhibit realistic attack behavior, which might include a decreased level or aggression or even retreating when suffering damage. Nonplayer Teammates: Given a squadron of soldiers, the group should move in a coordinated sup- portive manner. Such support NPCs are sometimes employed in multiplayer online games to assist inexperienced players. While in some contexts, this might be scripted by the game designer, typ- ically this is handled by the game’s AI system. Artificial Intelligence: Basics 76 CMSC 425
Image of page 76
Support and Autonomous Characters: This includes generating realistic crowd behavior, where the characters may need to interact in a realistic manner when coming into contact with the player’s character. Commentary/Instruction: Again, this is typically scripted, but an example requiring AI might involve determining whether the player is stuck and in need of a hint on how to proceed. The key element in all of these examples is the feature of complexity . Examples of things that are not AI include: Determined by physical laws: Examples include the way in which a basketball bounces off the rim of a basket, or the spinning motion of a car that just hit an obstacle. Purely random: For example, the shape of the next block that falls in a game of Tetris. Direct response to game rules/user inputs: This includes events for which the response is prede- termined by the game designer. This includes typical camera control, scripted animations, events that are triggered by the user’s inputs, and events that are scheduled to occur at a particular time or after a particular time delay. One notable gray area is where AI ends and animation begins. For example, a soccer player dribbling the ball must make decisions as to how to avoid opponents, which in turn affects the direction and speed with which he/she runs, which in turn affects joint angles. Typically, AI systems control the high-level decisions and the animation controls the lower level decisions: Should I run with the ball or pass it? Definitely an AI decision (unless scripted or determined by user input).
Image of page 77
Image of page 78

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture