On The Control Algorithm And Experimental Simulation Of Flocking Behavior For Swarm Robot System

Flocking behavior is a common phenomenon in nature, such as flocks of birds, schools of fish, and colonies of bacteria. It is found that flocking behavior makes the individuals of the swarm have certain advantages over single individual in many kinds of task, for example, avoiding predators, finding food, and completing some complex task which single individual can never achieved.Inspired by the flocking behavior in nature, the complex behavior of the swarm robot system is studied through local interaction between simple agents. Swarm robot system is an extend of multi-robot system with significant advantages: For one thing, it has good robustness, which means the swarm robots system can still complete the task even if some agents failed or disturbance emerges; For another thing it has strong adaptability, which means the swarm robots system have the ability to solve different tasks; and also it has extendibility, which means the system can move on normally with increase or decrease in the number of system agents.The research work on swarm robot system is of great importance both in theory and practice. It can help to reveal the essence of complex collective behavior in nature, and can imitate the intelligent behavior of biological beings perfectly. And with the improvement of manufacturing technology and decrease in cost for the development of science and technology, there will be wide application of swarm robot system in the future battlefield, the medical field, the aerospace and industrial field.The flocking behavior of swarm robot system is studied in this thesis. And the brief content of the thesis is as follows:Firstly, the fuzzy neighborhood between agents is described. The distance between agent i and its local neighbors are different, thus the effect of the control strategy of different neighbors are different on the agent. And it is necessary to have different intensities between agents. As fuzzy inputs, the distances between agents are utilized to obtain the intensities with neighbors through the fuzzy logic system. A control algorithm is designed to verify the fuzzy neighborhood of the swarm robot system. And the simulation results show that the swarm can achieve a stable structure and move synchronously. For further study of the robustness of the swarm robot system during the flocking behavior, simulations of the swarm system are performed with a failure agent, a communication failure agent and random disturbance, respectively. And the results demonstrate that the swarm has a good robustness using the controller put forward in this chapter.Secondly, the split mechanism for swarm robot system is designed under the environment of multiple virtual leaders. Taking the requirements of the leaders and the tracking conditions of the neighbors into account, the ant colony algorithm is applied to calculate the probability for agents to track the leaders and the roulette wheel method is used to determine the tracking target when there are multiple virtual leaders in the environment which need to be tracked separately. The attraction/ repulsion force between agents which have different leaders is deduced based on the force between agents which have the same leader. Under the effect of the attraction/ repulsion force, the swarm can split into different groups to follow different leaders. The simulation results reveal that the swarm can effectively split into different groups according to the leader's needs under the proposed split algorithm.Then, the controller is designed for the swarm robot system in an environment with unknown obstacles or damping factors. When there are obstacles (one or more) in the environment, the traditional artificial potential field method is improved to achieve obstacle avoidance based on the number advantage of the swarm. During the collision avoiding process, the agent get repulsion force from the obstacles, and approach to the center of the neighbors which did not perceived obstacles. To validate the proposed algorithm, experimental simulation is performed. And the results show that the algorithm can make the swarm avoid obstacles effectively. When there are unknown damping factors in the environment, the velocity of swarm can hardly reach the convergence. Therefore, the fuzzy adaptive method is introduced to adjust the online parameters of the controller. An algorithm is designed to simulate the situation and the results indicate that this method can decrease the influence of the unknown damping factors, making the swarm converge to the desired speed quickly.Finally, a novel communication strategy is proposed for the swarm robot system which has flocking behavior. As there is heavy communication traffic in the process that the swarm achieves collective behavior, particle swarm optimization algorithm is firstly used to get the best communicate frequency and reduce redundant communication. After that, the evaluation functions of the swarm structure and moving state are established. And different communication frequencies correspond to different evaluation values to further lighten the communication traffic. Experimental results demonstrate that the proposed strategy can reduce the communication while keeping the formed flocking behavior.In summary, four problems of the flocking behavior of swarm robot system are studied in this thesis. The research work focuses on the controller design for flocking behavior of swarm robot system under different external environment, and also a communicate strategy is proposed to reduce redundant communication. Simulation experiments are done to validate the proposed algorithm.