On The Cooperation Behavior Of Swarm Robots System

With the further research on biology, researchers find that some social insects can create complex intelligent collective behaviors through simple local interactions. These intelligent collective behaviors show good system level characteristics: robustness for individual failures, adaptability to environment changes and extendibility of colony size. Swarm robotics, which is a new research direction based on multidisciplinary cooperation, is right inspired by the collective behaviors of social insects. Swarm robotics is the further topic of the research on traditional multi-robot systems. Swarm robotics is the study of how to make large number of relatively simple robots emerge desired collective behaviors through the local interactions among individual robots and between the individual robots and the environment.From the 1990s, the research related to swarm robotics has already made some progress on theory and practice. More and more researchers from different disciplines engaged in the research on swarm robotics. The research on swarm robotics has great theoretical significance, which will promote the research on swarm self-organization theories and the laws for the emergence of cooperation behaviors. With the rapid development of microelectronic technology and the machining process, the design and manufacture technology for autonomous robot will be improved continuously, and the application of swarm robots system to industry, physic, military, spaceflight and other fields will be more extensively. The primary task is to make the cooperation behavior satisfy the system demand, while using swarm robotics to solve problems. So, it is particularly important to do research on the cooperation behavior.This dissertation studies the cooperation behavior of swarm robots system, based on the tasks of swarm robots foraging and formation control. The major research work of this dissertation is composed of the following three aspects.1. The distributed aggregation formation control strategy is studied. The swarm robots systems run high risk of collision or interference when executing task, so it is necessary to carry out efficient motion coordination. This dissertation studies how to make swarm robots form and maintain aggregation formation while keeping certain intervals.Firstly, based on Boid model, combined with behavior based motion control, the self-organized aggregation formation control strategy for swarm robots is presented. The aggregation formation control strategy introduces target approaching behavior and obstacle avoidance behavior, in order to satisfy the need of practical application. Simulation experiments verified that the swarm robots can form and maintain aggregation formation, avoid the unknown obstacles in the environment and approach target area.The shape of the aggregation formation, which is uncertain based on Boid model, is usually expected to be predictable and unchanged when swarm robots execute certain tasks. Therefore, the aggregation formation control strategy based on social potential fields is proposed, which make swarm robots form and maintain near circle aggregation formation. The thresholds are set to improve the speed of approaching target area, flexible social potential fields mechanism is proposed to improve the speed of obstacle avoidance, simulation experiments data indicates the modified aggregation formation control strategy improves the operation efficiency considerably. The aggregation formation control strategy achieves the coordination of aggregation behavior through local perception of individual robot, and thus has good scalability and robustness.2. Distributed task allocation strategy is designed. Task allocation is an essential mechanism for swarm robots to create efficient cooperation behavior. The primary goal for swarm robots task allocation is to realize dynamic labor regulation through local interactions. Swarm robots task allocation is fully studied in this dissertation, and the foraging mission which aims at maintaining food consumption is selected as the research background.Self-organized labor regulation mechanism for social insects is referred, and the task allocation strategy based on response threshold model is designed: Individual robot firstly calculates state switch probability according to the intensity of the stimulus associated with specific task, and then the foraging behavior is regulated utilizing the probability calculated. Response threshold for individual robot is unchanged, while the response curve gradients are different. This will make individual robots response differently to the same stimuli. Simulation experiments show that the task allocation strategy can satisfy food consumption and effectively achieve labor division for swarm robots system. The task allocation strategy can adjust the labor division accordingly when the total quantity of food in the foraging scenario is changed or the food consumption rate is changed.This task allocation strategy has good adaptability, and multi-foods consumptions can be satisfied by setting one response threshold for each kind of food. Local communication mechanism is designed in order to improve foraging efficiency. Robots can get useful foods information through local communication. Comparison experiments verified that the local communication mechanism can effectively improve the performance of the task allocation strategy.3. The collective cooperation behaviors are analyzed using mathematical models. Mathematical model provides an effective method for studying swarm robots system. By establishing the mathematical model for swarm robots system, the dynamic characteristics can be analyzed, the collective behaviors can be predicted and the parameter optimality can be verified. This dissertation establishes the mathematical models for swarm robots systems and analyzes the collective cooperation behaviors using these mathematical models.Firstly, main states transitions for the colony are generalized, then the equations depicting the variation of quantities of individuals in different states are established according to rate equation, and the mathematical model for swarm robots system can finally be obtained by establishing simultaneous equations. The modeling and analyzing results for swarm robots cooperative foraging indicate that the efficiency of cooperative foraging will firstly increase and then decrease with the increase of waiting time, namely there is an optimal waiting time which can maximize the efficiency of cooperative foraging; simulation experiments show that the analysis results of the mathematical model agree properly with the data obtained from the simulation experiments. The quantitative analysis of the mathematical model for heterogeneous swarm robots foraging indicates that the effect of interference will reduce the improvement on the finish time of foraging task when increasing the number of outer robots, and the improvement on the finish time of foraging will decrease when increasing the rate for inner robot to find food. For the foraging based on division mechanism, the modeling and analyzing results indicate that the efficiency of foraging based on homogeneous strategy is higher than that based on division strategy. Mathematical model provides an effective means for analyzing and predicting the collective behaviors, and has great significance for designing and improving control strategy for swarm robotics.In summary, methods and theories for swarm robotics are studied in this dissertation. The primary goal of this work is the feasibility and innovation of the methods for the emergence of cooperation behaviors for swarm robotics. Simulation experiments are carried out and mathematical models are established for the purpose of related verification and analysis.