The Artifical Intelligence In Racing Game

The game industry has a rapid development in recent years, the performance of intelligence in games relative to the near-ultimate image quality has became a bottleneck restricting the development of the game, this contradiction in the racing game is especially prominent. The intelligent performance of the rival against the player in racing game is the key to quality of the game. Skillfully through the intelligent control of vehicle automatic driving is not only the problem which racing game need to address, but also has great practical significance to the robot simulation of reality. This paper aims at the problem of how to better control car racing, studied the artificial intelligence applications in the racing games. Main tasks include the following.First, In this paper, the concept of intelligent game, features and applications are described in detail, research, and summarizes the application of artificial intelligence game situation, on the racing game, the status of applications of artificial intelligence have been classified in research on artificial intelligence methods in competition speed in the game and the future applications of that in the racing game how to better control the car to meet the complex environment of the scene, and generate reasonable and effective behavior has become an urgent problem to be solved.Subsequently, this paper aim at the problem how to control the car around a real physical path of connectivity problem, present three car control mechanisms based on evolutionary algorithm and neural network. The basic idea of the control mechanisms is to use the feature of optimization algorithms, get the optimized result behavior with some particular information inputs to achieve the purpose of better control of the car. This paper studied that what information needs to control the car around the path skilled and quickly, how the information be expressed and how to evolve a car controller around the track as quick as possible through comparing different controllers in virtual environment experiments. Experiments show that the behavior generated by sensor controller achieved good results, and its fitness is better than other controller. It also shows that the first person information of the car has more advantages than the third person information.Then, this paper studied the issues that how the car controller to simulate the behavior of player behavior, designed a multi-objective evolutionary strategy of total progress, the speed and direction offset of the car passing through the way points, through comparing the direct and indirect simulations to supervised train the controller to simulate the behavioral characteristics of players. Experiments show that the indirect simulation has a good performance of the three evaluation criteria, and simulate the player driving style.Finally, this paper achieve a car racing game based on Torque game Engine, which is the competition of the player and the NPC cars to test the feasibility and practicality of the earlier application. The game developed shows good gameplay.