| Swarm robotics is a method to study how to control coordinately a large number of robots;it is an application of swarm intelligence in a multi-robot system.A swarm robot system,based on the coordination of many simple individuals,can complete complex tasks that cannot be completed by a single robot.The design cost of a swarm robot system composed of simple individuals is also generally lower than a single robot with equivalent capabilities.The research into swarm robots has received extensive attention due to their characteristics of robustness,flexibility,and scalability.Further,swarm robot foraging is a typical task in the research of swarm robots.It is a comprehensive process that integrates task allocation,area exploration,grabbing,and transmission.The research of swarm robot foraging is of great significance in both theoretical research and practical application.In terms of practical applications,swarm robot foraging has broad application prospects,such as toxic waste cleaning,sample collection in unknown areas,mine clearance,search and rescue,and planetary detection.In theory,swarm robot foraging can promote the understanding of the emergence mechanism and evolution law of swarm self-organization behavior;and therefore realize the design,analysis and control of a swarm robot self-organized behavior.Area exploration and task allocation are two important branches of the swarm robot foraging research;their algorithm parameters are key factors to the performance of the algorithm.Based on basic mathematical theories and machine learning theories,this dissertation has conducted in-depth research on area exploration,task allocation strategies and parameter optimization strategy in task allocation algorithms.The specific contents are as follows:(1)Due to limited individual abilities of the swarm robots,namely local sensing and low processing power,random searching becomes the main search strategy used in swarm robots.However,duplicate searches can significantly drag down the efficiency of random searching.By analyzing the characteristics of swarm robotic exploration,a random walk method based on adaptive step lengths is proposed,that is,each robot adjusts its step length adaptively by estimating the density of robots in the environment.By doing this,the number of duplicate searches is reduced.The effectiveness of the proposed method is testified using simulation and robot experiments,and the performance of the proposed method is also accessed when applied to area search tasks.The experimental results show that the searching efficiency of this method is significantly better than other methods.(2)Although the random walk method has been widely used in area exploration,the existing research has paid little attention to the factors affecting the searching efficiency.Firstly,the formula of the mean square displacement of the robot position is given,and it is conducted that the mean and the variance of the random step length are the factors that affect the searching efficiency.In addition,a truncated random walk method is constructed to make the generated step lengths follow a given distribution and the step lengths are within a specified range,thereby improving the searching efficiency.Lastly,the correlations between the step length threshold,the area of the region,and the number of robots are provided based on mean square displacement and truncated random walk method.When the expectation of the step length is greater than the step length threshold,the swarm robots can achieve the highest searching efficiency.The experimental results have proven the effectiveness and correctness of the method.(3)Algorithm parameters have a relatively great impact on the algorithm performance.It is an important task to optimize the parameters.The artificial bee colony algorithm has a wide applicability in parameter optimization due to its simplicity and convenience for implementation.The solution search equation in artificial bee colony algorithm has a relatively high exploration ability,but its exploitation ability is poor.To overcome the limitations of the existing methods,five solution search equations are used and a self-selection mechanism based on the iteration success rate is constructed to improve the optimization performance of the algorithm.On the benchmark test set,the performance of this method is far superior to the existing methods.Finally,the proposed artificial bee colony algorithm based on a self-selection mechanism is used to optimize the parameters involved in the task allocation algorithm.(4)In swarm robot foraging,task allocation refers to the process of dynamically adjusting the number of foraging robots based on task requirements and changes in the environment.Because of a lack of consideration in existing methods about the physical interaction between robots and the traffic condition in the search area,the method that uses the traffic flow density and the number of obstacle avoiding occurrences between robots to measure the traffic conditions in the environment is proposed.Based on this,a dynamic threshold is constructed.Then,a response threshold model based on the Sigmoid function is constructed to calculate the robot's foraging probability.The experimental results show that this method can not only improve the foraging efficiency but also reduce the number of physical interactions between robots.(5)The existing task allocation methods are over-sensitive to the external environment and noise,and their robustness is poor,while the robustness of the system is crucial to swarm robot foraging tasks.A simple and robust attractor selection model has been used to control the robot's motion.This model can make a robust response to changes in the environment without multiple,complex sensors.Thereafter,the traffic flow density and the number of obstacle avoiding occurrences between robots are used to adjust the threshold.Then,a dynamic response threshold model is constructed to calculate the robot's foraging probability,and to achieve task allocation.The experimental results have proven the effectiveness of the method. |
PDF Full Text Request