Research On The Pursuit Of Large-scale Emotional Robot

In recent years,the research of artificial emotion theory is becoming more and more mature.The emotion is introduced into the multi robot system,which makes the robot more intelligent and autonomous.At the same time,it can play an emotional role in task decision and further improve the performance of multi robot system.As a typical multi-robot system,the pursuit of emotional robot team has become one of the hotspots in the research of the emotional robot system.The research of emotional robot team pursuit is mainly divided into two aspects:task allocation and pursuit strategy.Through the improvement of algorithm,the completion time of pursuit task is the shortest or the benefit is the biggest.This dissertation focuses on the large-scale emotional robot team.By studying the algorithm of pursuit task allocation and pursuit strategy,we effectively play the advantage of emotion in large-scale robot pursuit.The main contents are as follows.(1)This dissertation introduces the model of emotion,mood and personality,as well as the relationship model of personality-mood and mood-emotion.At the same time,the emotion cooperation factor,which is the important reference basis of the emotion,is changed by emotional attenuation and external stimulus.We also analyse the characteristics of large-scale emotional robot pursuit and the influence of emotion on the pursuit results.(2)This dissertation proposes an algorithm based on self-organizing algorithm for the task allocation of emotional robot pursuit.The algorithm combines the self-organizing algorithm of SOM neural network,and integrates the emotional robot task allocation into the training process of network.So,it reduces the time complexity of task allocation for large-scale emotional robot pursuit team.According to the actual factors such as emotional cooperation factors and distance,emotional robots generate the value of competitive winning function to judge whether it is suitable for a pursuit team.In addition,we use reinforcement learning to adjust the value of competitive winning function,which optimizes the role of emotion in task allocation.The experimental results show that the proposed algorithm is suitable for the task allocation of large-scale emotional robot pursuit and shortens the total pursuit time.(3)Combine with the task allocation algorithm proposed in the previous thesis,we propose the algorithm for multi emotional robot pursuit based on TMSM-ANN potential distribution.After the task allocation algorithm,the algorithm sets up the pursuit potential points for the evader in the team.The team's pursuers generate emotional willingness and distance to all potential points.Then,these parameters input ANN to allocate potential point for each pursuer.The pursuer uses artificial potential field to catch the evader.In order to converge to the global optimal weight faster,we use TMSM artificial immune algorithm to optimize the weight of the ANN.The set of potential points enables the pursuer to ambush and encircled the evader,and takes advantage of the emotional selection in process of pursuit.The effectiveness of the proposed algorithm is verified by simulation experiments.Combining with the previous task allocation algorithm,the total pursuit time is optimized.