贝叶斯网络与模糊理论的移动机器人避障  

Obstacle Avoidance for Mobile Robot Based on Bayesian Networks Model and Fuzzy Theory

作　　者：柴慧敏[1] 陈奋增 方敏[1] 赵昀瑶 CHAI Hui-min;CHEN Fen-zeng;FANG Min;ZHAO Yun-yao(School of Computer Science and Technology,Xidian University,Xi’an 710071,China)

机构地区：[1]西安电子科技大学计算机科学与技术学院,陕西西安710071

出　　处：《控制工程》2020年第10期1657-1664,共8页Control Engineering of China

基　　金：陕西省工业科技攻关项目(2016GY-112)。

摘　　要：针对未知环境下的移动机器人避障问题,提出了贝叶斯网络模型和模糊理论相结合的避障方法。首先根据机器人避障的基本情况分类,建立了避障贝叶斯网络模型;在避障过程中,机器人行进的前方[10°, 170°]范围划分为5个区域,依据传感器数据提取不同区域的障碍物目标要素;分别建立障碍物目标的距离和角度属性的模糊隶属函数,将障碍物目标要素的距离和角度信息映射到避障贝叶斯网络模型征兆节点的取值状态上,以获得推理的输入证据;采用贝叶斯网络消息传播的推理方法,计算结果节点的后验概率值,选择后验概率值最大的取值状态,作为移动机器人避障的动作处理结果。通过对避障贝叶斯网络模型的实验测试,及在4种典型仿真场景下的避障测试,结果表明:避障贝叶斯网络模型的推理结果与预期相符,机器人能够安全通过障碍区。For the obstacle avoidance of mobile robot in unknown environment, an obstacle avoidance approach based on Bayesian networks model and fuzzy theory is proposed. Bayesian networks model for obstacle avoidance is constructed. The angle rang [10°, 170°] in the front of mobile robot is divided into five parts. According to the sensor data, obstacle target factor of the five parts can be extracted. Fuzzy member functions for distance and angle of obstacle target are provided respectively. By these functions, the information of distance and angle can be mapped into the corresponding states of symptom nodes in Bayesian networks model and the input evidences of inference are obtained. Using message-passing algorithm, the posterior probability of outcome node can be calculated. The state which posterior probability is maximum is chosen for the result of obstacle avoidance. Experiments for Bayesian networks model for obstacle avoidance and four typical simulation scenes are given. The results of tests show that the inference results of Bayesian networks model for obstacle avoidance accord with expectations and mobile robot can pass the obstacle area safely.

关 键 词：移动机器人 避障 贝叶斯网络 模糊理论 消息传播推理 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]