论文部分内容阅读
本文提出了一种用于自主式移动机器人的障碍物类型识别的数据融合新方法，有两种不同的神经网络——小脑模型联接控制器（ＣＭＡＣ）和多层前向网分别对来自ＣＣＤ摄象机的二维图象和来自超声测距系统的距离信息进行数据融合，而这两种神经网络事先都用围绕障碍物采集的数据集进行过离线训练．为了验证该系统的有效性，我们构造了一系列的仿真实验．实验结果表明，一台个人计算机就能实时地识别出障碍物的类型．
This paper presents a new method for data fusion of obstacle type recognition for autonomous mobile robots. Two different neural networks, the CMAC and the multi-layer forward network, The two-dimensional image of the camera and the distance information from the ultrasound ranging system are used for data fusion, and both neural networks are previously trained offline with the data sets surrounding the obstacle acquisition. In order to verify the effectiveness of the system, we constructed a series of simulation experiments. Experimental results show that a personal computer can identify the type of obstacle in real time.