An AI based solution for the control of 3D real-time sensor based gaming

Abstract

Current approaches to game strategy make reference to the use of AI techniques in order the better predict the players' behavior. More recently the 3D sensor based gaming has spread rapidly in the PC and console gaming and gamers' communities. The sensor based gaming has changed dramatically the gaming design forcing it to move away from triggered scripts to more interactive reactions. This requires that games are designed such that they can adapt faster to the gamer reactions and behavior forcing the characters to react in an intelligent way by analyzing a variety of options and moves made by the user in the real space. The advent of 3D cameras and of the augmented virtual reality insertions into the game space via external sensors detecting users' moves and not buttons or keys pressed requires considering a new way of building a computer game. There is a stringent need to produce more and more intelligent engines to drive the game. In this paper we introduce and discus a series of aspects of game building when a 3D camera and movement sensors are used. The AI engine composition is discussed and shown that it has at its basis the opponent model which helps the engine learn from the data acquired. An evaluation function is also necessary. This evaluation function is built and tested on real data. The AI engine is also investigated in relation to a special language based on body or body parts movements.