Intelligent Perception Of The Virtual Characters To Achieve The Routing Study

Along with the robot technology's development and large-scale game's universal, the NPC's AI (Artificial Intelligence) is getting more and more important. Its application's universality and the high hierarchical obtain more and more attention. In the current AI technology, NPC, the main body of AI, is still dull and takes orders from programmer's arrangement, instead of acting through oneself"ponders". NPC does not have eyes and other feelings, gets the information needed completely through the programmer. Such AI system is getting bigger and bigger in the map, the NPC integer are getting more and more, however, the environment is getting more and more complex in the new generation games, such AI system could not meet its need already.This paper proposed that one kind of cognition, the decision-making, the execution,three advancement's-like sensations and march forward the system,lets NPC use his own thinking mode understanding environment,makes the decision-making with their own judgment and carries out. In such an AI system,the programmer could control NPC through controlling his thinking mode instead of controlling trivially each NPC's detailed behavior. Therefore,this paper first discussed the topic research background and this method in the domestic and foreign research present situation.To the AI system's research,this paper proposed separately the screw traversal algorithm and the beam traversal algorithm,and also optimized precomputed search tree algorithm.In the sensation algorithm's research aspect, this paper proposed the screw traversal algorithm and the beam traversal algorithm, processing environment's physics and geometry data to the logical data which NPC can understand. The screw traversal algorithm is by presses the spiral direction in view of its central point traversal algorithm from near to far. The beam traversal algorithm basis beam's geometric property deals with the line of sight issue in the present paper the traversal algorithm. This two algorithms in the environment sensation, in view of the convex polyhedron obstacle, can when marks the grid, can realize the fast traversal, and reduces the environmental information to a great extent the repeated transfer. Through the cognition, NPC may select many appropriate essential regions, and transmits its target point for diameter seeking system in order to complete the following implement process.In diameter seeking algorithm research aspect, this paper optimizes the goal-driven precomputered tree diameter seeking algorithm. Goal-driven precomputered tree diameter seeking algorithm is a real-time, overall importance diameter seeking algorithm, and may realize partial evades bonds the function. This paper used the complete anticipated tree algorithm replaced A* algorithm in the original reference, enhanced anticipated number's efficiency, reduced the algorithm average working time, thus improved the efficiency of the whole seeking diameter algorithm, and has conducted the research to the beginning flexibility, caused to seek the diameter algorithm and the sensation algorithm even more agrees with. In addition, the NPC movement conducted through anticipated way and uses the limited automatic state machine (FSM) to carry on the storage and the transfer.Finally, this paper summarized the full text work and proposed the next step might carry on work.