Bots with Attitude

Abstract

Bot navigation in many computer games is achieved with static waypoints placed throughout a map. This allows pathfinding to be performed by implementing search algorithms to find the optimal path between waypoints using a heuristic function. This heuristic is usually based on Euclidian distance and terrain. While this type of pathfinding is effective, it does not provide for adaptable pathfinding in a dynamic environment. One concept that promises to achieve adaptive responses in differing situations in artificial agents is that of emotion. In this paper, we examine the use of positive and negative feelings, referred to as attitude, in developing a dynamic heuristic for A* that responds by finding paths between waypoints influenced by a bot's internal state and its feelings toward objects in the game. Using a traditional A* heuristic, a bot needing to travel to two goal locations will firstly perform pathfinding to move to the first goal followed by pathfinding for the second goal. Using our dynamic heuristic, the second pathfinding operation is eliminated as a path to the second goal via the first goal is calculated. This dynamic heuristic is illustrated in this paper through experimentation with a bot traversing a map of random waypoints.