Charles, Darryl. "Enhancing gameplay: challenges for artificial intelligence in digital games." DiGRA Conference. 2003.
Keywords: Artificial Intelligence, dynamic learning, gameplay
What’s the Big Idea: The implication of enhanced utilization of AI in games and gameplay are discussed
Terms to get familiar with: dynamic learning, perceptron [21], adaptive genetic algorithms, recursive neural networks
Research questions/Hypothesis, methods used in research/conclusions: Existing games are evaluated. It’s interesting to note that a player must believe that intelligent behavior is being exhibited, otherwise any AI coding is much less effective. Also interesting to note: the disparity between the amount of effort required to create effective AI and the gains that are clearly visible and accessible to the player is one of the main reasons why the use of AI has generally stabilized.
Techniques typically used in older games: finite state machines for character and object behavior. Path finding techniques such as the A* algorithm for character and vehicle movement. Later games like Quake have genetic algorithms and neural networks to train “bots” off-=line so they will have enhanced capability in the game. The Sims is probably the first game to use “intelligent objects” that pass information to each other about status, such as a fridge telling a passing Sim it has food. Not a lot of ground breaking AI innovation in commercial digital games.
Half-life’s AI strong point is co-operative opponent behavior makes players consider strategy carefully. The opponent seems to have an intelligence plan, adapting on the basis of player behavior, making the play more rewarding and interesting challenge.
Improved AI increases the degree of immersion in a game.
Areas of innovation include storytelling, dynamic learning, affecting emotion.
Dynamic learning is learning in-game, as opposed to training bots out of the game. Potential gain with dynamic adaptation to player behavior, play patterns and skill level.
Research Papers to read from this article:
21. Rosenblatt F, “Principles of Neurodynamics”. Spartan Books, 1962
23. Wen Z. et al, “Neural Networks for Animating Variations in Character Behaviours”
”GAME-ON 2002, 3rd International Conference on Intelligent Games and
Simulation, pp. 189-196.
Keywords: Artificial Intelligence, dynamic learning, gameplay
What’s the Big Idea: The implication of enhanced utilization of AI in games and gameplay are discussed
Terms to get familiar with: dynamic learning, perceptron [21], adaptive genetic algorithms, recursive neural networks
Research questions/Hypothesis, methods used in research/conclusions: Existing games are evaluated. It’s interesting to note that a player must believe that intelligent behavior is being exhibited, otherwise any AI coding is much less effective. Also interesting to note: the disparity between the amount of effort required to create effective AI and the gains that are clearly visible and accessible to the player is one of the main reasons why the use of AI has generally stabilized.
Techniques typically used in older games: finite state machines for character and object behavior. Path finding techniques such as the A* algorithm for character and vehicle movement. Later games like Quake have genetic algorithms and neural networks to train “bots” off-=line so they will have enhanced capability in the game. The Sims is probably the first game to use “intelligent objects” that pass information to each other about status, such as a fridge telling a passing Sim it has food. Not a lot of ground breaking AI innovation in commercial digital games.
Half-life’s AI strong point is co-operative opponent behavior makes players consider strategy carefully. The opponent seems to have an intelligence plan, adapting on the basis of player behavior, making the play more rewarding and interesting challenge.
Improved AI increases the degree of immersion in a game.
Areas of innovation include storytelling, dynamic learning, affecting emotion.
Dynamic learning is learning in-game, as opposed to training bots out of the game. Potential gain with dynamic adaptation to player behavior, play patterns and skill level.
Research Papers to read from this article:
21. Rosenblatt F, “Principles of Neurodynamics”. Spartan Books, 1962
23. Wen Z. et al, “Neural Networks for Animating Variations in Character Behaviours”
”GAME-ON 2002, 3rd International Conference on Intelligent Games and
Simulation, pp. 189-196.