Research Of Dynamic Multi-robot Pursuit Algorithm Based On Jerne Immune Network

Studying on multi-robot dynamic pursuit system has very important significance. This paper studying on the pursuit system is a group of autonomous mobile robots through mutual cooperation to effectively capture another group of robots in unknown environments. Its essence is the study of multiple autonomous mobile robot composing the optimal team under what kind of system structure and pursuing multiple moving targets by effective collaboration strategy.The immune system is a highly parallel distributed processing system, it has distributed, adaptive, dynamic balance, fast, robustness and other features that the multi-robot pursuit system is needed. In view of this, this paper studys the algorithm of multi-robot dynamic pursuit system based on the work mechanism and characteristics of the Jerne immune network. Details are as follows:(1) This paper summarizes the research status of multi-robot system and multi-robot pursuit problem, introduces the main contents of multi-robot dynamic pursuit system, including system architecture, task allocation and pursuit strategy.(2) In the study of the theoretical basis for multi-robot pursuit systems-the concept of the human immune system, composition, structure, function, features, Jerne’s idiotype immune network theory; and the concept of the artificial immune system, studying content and status in the field of robots; proposed groups and individuals architecture of multi-robot system. With the groups hierarchical architecture, robots are divided into different task teams autonomously in the face of dynamic targets, reducing the difficulty of coordination between the robots to meet the real-time, dynamic adaptability and robustness. With the individual deliberative/reactive hybrid architecture, highlighting the robot’s ability of real-time response, autonomous decision and coordination.(3) Based on the above architecture, this paper studies the robot’s behavior decision-making module, proposes a multi-robot cooperation detection algorithm based on Jerne immune network. The algorithm leads into the role of T cell based on B cell immune network model, establishes a detection algorithm based on BT cell immune network formula, solves the problem of randomly selected part detection points existing in B cell immune network model detection algorithm, achieves traversal detection while reduces duplication detection rate and improves the detection efficiency of the robot.(4)In this paper the pursuit issue is break down into multiple levels to study and proposes an algorithm for each level. Under the combined effect of these algorithms, improving the rapidity, flexibility, coordination of the pursuit system and ability to adapt to the dynamic environment. Each algorithm’s functions as follows: immune anxiety algorithm, introducing the concept of anxiety which the robot accept the task invitation in the opportunity decision-making, solving the problem of the task allocation algorithm based on contract net that the robot is forced to accept the task invitation in inappropriate timing. Immune decision algorithm, immune contract algorithm, using the role of antigen and interaction of antibodies in Jerne immune network theory, achieve a multi-robot task assignment. Immune pursuit algorithm accelerates the speed of the pursuit team to perform task. And, in the design for the pursuit system it reduces the traffic by setting the state of the robot, ensures rapid response capability for multi-robot pursuit system.(5)Based on Georgia University and Carnegie Mellon University developing TeamBots simulation platform, combining own research issue to develop a practical, more powerful Java classes, the newly developed classes obey TeamBots developing rules,it is seamlessly connection with the simulation platform, extends the functionality of the simulation platform, provides a standard test platform for in-depth studying on the multi-robot pursuit target problem. The simulation system is modular in design, easy to carry out simulation studies for the new algorithms, and immune pursuit algorithm mentioned in this paper were verified. The research results show, in the implementation of an unknown environment exploration mission, the multi-robot dynamic pursuit system based on Jerne immune network can solve the randomness of detection algorithm based on B cells immune network formula, achieve a traversal detection, reduce duplication detection rate and improve the detection efficiency of the robot. In the pursuit task execution, the system can quickly and efficiently complete the pursuit task; and traffic and computation between robots is smaller that can be applied to a large number of multi-robot system, the entire system reflects better distributed, adaptive, homeostasis, robustness and other characteristics. Results of this study provide a new ideas for the study of multi-robot pursuit system, has important theoretical and practical significance for studying the environment detection and pursuit in unknown environment by Jerne immune network theory.