Game AI Analysis - Movement
Learning:
Artificial Intelligence - Movement:
Computer Graphics (CG) Modelling: addressed automation (and specifically movement) in games. Ranging from the early geometric (physicality), to the more recent challenges of behavioural models; in which self-animating characters, react appropriately to perceived environmental stimuli. However, animations still need programming, and (Millington & Funge, 2009) agree that movement AI refers to algorithms that turn decisions making into any form of motion.
“Cognitive models: go beyond behavioural models, in that they govern what a character knows, how that knowledge is acquired, and how it can be used to plan actions. Cognitive models are applicable to instructing the new breed of highly autonomous, quasi-intelligent characters that are beginning to find use in interactive computer games. Moreover, cognitive models can play subsidiary roles in controlling cinematography and lighting”, (Gamasutra, 1999).
Behavioural: how should an object or artefact in a game behave in situations; built upon lower layers.
Physical: Mass, additional forces and collision; considers the forces that Kinematic mechanics do not.
Kinematic: mechanic that describes the motion of points, bodies (objects), and systems of bodies (groups of objects); while not considering the mass of each point, body of system of bodies or the forces that caused the motion. Can often be referred to as the "geometry of motion" and is occasionally seen as a branch of mathematics. Builds upon Geometric models.
Geometric: Considers the form and shape of an object; made from Vertices Transformations, edges and faces.
Movement: tends to be a way of displaying behaviour in games and typically takes the form of an animation, and generally reflects the hidden processing of the AI to the player.
AI techniques:
Behaviour & Movement AI:
Action selection: has the strength of simplicity, due to a variable or state defining the display and use of action (movement), that will take place. Essentially Action Selection, is just a Boolean or set of if statements or a switch case controlling actions.
BIODs: which have the strength of emulating the pack movement of prey and predator; and the additional strength of using simple spherical colliders for sensing and typically follow three methods, these are:
1. Separation - to prevent overcrowding.
2. Cohesion - moving towards and average position.
3. Alignment - steer towards an average heading.
IT is key to note, that movement and steering are not the same thing, movement is a velocity to move forwards to a destination, and steering is to recalculate the velocity, to avoid collision and objects and change course.
Pathfinding: typically involves a grid or node-based system, that has an array of costs associated with each grid position or node. Two positions tend to be used in path finding, the position you currently are in and the desired target position. A strength of the algorithm is that it makes a series of probes into other nodes recording the cost until a path of nodes have been selected that reach the destination node; improvements to the algorithm find both shortest route, but also the cheapest in terms of cost; examples being A* and Dijkstra. However, a weakness of the algorithm is that it will require more memory and processing resources than other game AI techniques.
Reflection:
Behaviour AI models build upon the lower layers of the modelling pyramid, and encapsulate movement AI. Movement in games, is more than just moving from A to B; movement itself is a reflection of what the gaming artefact represents to the player and must correlate with players mental models. However, movement can also give indication, as to what to expect from the behaviour AI. However, AI is moving beyond behaviour AI models to Cognitive Ai models which encapsulate Behaviour models of AI - moving beyond behaviour, to what is the inner state of the AI – i.e. what is the AI thinking?
References
Gamasutra, 1999. AI for Games and Animation: A Cognitive Modeling Approach. [Online] Available at: https://www.gamasutra.com/view/feature/131812/ai_for_games_and_animation_a_.php [Accessed 12 February 2018].
Millington, I. & Funge, J., 2009. Artificial Intelligence: FOR GAMES. 2nd ed. Burlington: Morgan Kaufmann.
