基于改进机器学习的图书馆机器人自主避障控制研究

Research on Autonomous Obstacle Avoidance Control of Library Robots Based on Improved Machine Learning

为控制图书馆机器人在行进过程中自动躲避障碍,达到理想工作效果,提出基于改进机器学习的图书馆机器人自主避障控制方法;采集图书馆机器人与目标障碍物距离信息,感知环境特征向量,当成卷积神经网络输入,经卷积、池化等操作,输出图书馆机器人对当前环境感知结果,该结果经输入输出变量模糊化、模糊推理以及输出变量解模糊等操作后,实现图书馆机器人自主避障无冲突运行;实验结果表明:该方法自主避障控制效果较好,避障行驶距离短,高速运行时反应更快,能够避开多个障碍物,识别分类结果与实际感知环境类型一致。

In order to control library robots to avoid obstacles automatically in the process of traveling and achieve an ideal working effect,an autonomous obstacle avoidance control method of library robots based on improved machine learning is proposed.Collect the distance information between the library robot and the target obstacle,perceive the environment feature vector taken as the input of convolutional neural network,and output the perception results of library robots in the current environment after the convolution and pooling.The results are processed through operations such as the input and output variable fuzzification,fuzzy reasoning,and output variable defuzzification,thus implementing autonomous obstacle avoidance and non conflict operation of library robots.Experimental results show that this method has the advantages of good autonomous obstacle avoidance control effect,short obstacle avoidance driving distance,and faster response when running at high speed.Meanwhile,it can avoid multiple obstacles,and the recognition and classification results are consistent with the actual perceived environmental types.