Research Of Multi Robots Capturing Strategy Based On Fuzzy Inference System

With the continuous development of robotics, the bottleneck of single robot has gradually appeared in structural design, perception, and solving complex tasks etc. Because the multi robot system has significant advantages in many aspects, it has been paid much attention by domestic and foreign scholars. As a typical task of multi robot system, multi robot hunting task often used to evaluate the performance of multi robot system which reflects, many characteristics of multi robot cooperation.Multi robot hunting task is defined as a process that multiple robots round up a single robot cooperatively by using its onboard sensor in continuous environment. In this paper, for multi robot hunting, a two-layered fuzzy inference system which is used for capturing was established, and two kinds of the multi robot hunting strategies which is base on fuzzy inference system, is also discussed. The first layer of inference system is for strategy decision which separates the hunting task into three stages: searching, approaching and hunting. Then executing different stages based on different strategies. According to the hunting stage, two strategies are discussed. One is the angle-controlled hunting strategy based on fuzzy inference system. The other one is the multi robot hunting strategy by using the fuzzy knowledge rule base from artificial immunity.The first strategy from a single robot motion control perspective, establish multiple angle lines based on strategy. According to the same set of inference system, each robot approaching the target in diagonal method, then close to escapees.In considering the motion of each robot in the multi robot system for completing a specific task which reflects a strong correlation, from the point of overall control of multi robot multi robot hunting strategy by using the fuzzy knowledge rule base from artificial immunity is discussed. First, analyzing related environmental factors, that is used for describing the hunt state which is the antecedent of the fuzzy inference system. The output is designed for the hunters the whole movement behavior. However according to the fuzzy inference system design above including huge input and output factors, that leads to large amount of fuzzy rules, which makes it difficult to Formulate directly through the experience. Therefore, the artificial immunity algorithm is used to obtain fuzzy rules. The individual is coded. A large number of individual form the population. Affinity function evaluate the ability. Then after a series of artificial immunity operation, get the most optimal fuzzy rule base. These rules are applied to multi robot hunting task for implementing the multi robot hunting.According to the actual environment, a simulated experiment environment is generated in this paper, and all the robots are ensured in accordance with the kinematic model of wheeled robot. The above two algorithms in different environments by simulation and analysis, the result shows that the methods above can effectively complete the multi robot hunting tasks, which leads to the ultimate hunting.