基于神经网络的移动机器人室内定位方法

控制工程 ›› 2022, Vol. 29 ›› Issue (12) : 2329-2335.

基于神经网络的移动机器人室内定位方法

作者信息 +

Indoor Localization of Mobile Robot Based on Neural Network

Author information +

文章历史 +

摘要

为了提高餐厅、商场、医院等半结构化室内环境中移动机器人定位的实时性与准确度，提出了一种基于神经网络融合编码器、激光雷达的定位方法。利用采集到的编码器数据通过航迹推算得到机器人的里程计数据，然后通过卡尔曼预测和拉格朗日插值的方式对里程计数据、激光雷达的数据进行数据增强，将增强后的数据作为一个三层神经网络的输入特征。最后，在 Turtlebot 机器人实验平台上进行实验。实验结果表明，所提定位方法提高了定位精度，具有良好的实时性，并且能有效应对环境干扰。

Abstract

In order to improve the real-time performance and accuracy of mobile robot localization in semi-structured indoor environment such as restaurants, shopping malls and hospitals, a localization method based on neural network combined with encoder and laser radar is proposed in this paper. The collected encoder data is used to calculate the odometry data of the robot by dead reckoning. Then, the odometry data and the radar data are augmentated by Kalman prediction and Lagrange interpolation, and the augmentated data are used as the input features of a three-layer neural network. Finally, the experiment is carried out on the experimental platform of Turtlebot robot. The experimental results show that the proposed localization method improves the localization accuracy, has good real-time performance, and can effectively deal with environmental disturbances.