多层感知器自监督在线修正的道路识别算法

为解决自主移动机器人非结构化道路识别检测准确性、鲁棒性及实时性的问题，提出一种基于感兴趣区域(Region of Interest，ROI)与多层感知器(Multi-Layer Perceptron，MLP)为核心的自监督在线修正算法.首先，通过ROI算法规定被处理图像的有效计算区域；其次，利用多层感知器对样本数据进行训练，将感兴趣区域按相应特征实现分类处理，并对分类区域进行形态学处理及特征提取处理，筛选出有效的行驶区域；最后，通过自监督在线修正算法替换错误处理结果，进一步保障道路分类识别的准确性.实验结果表明，改进算法能准确地识别出环境中的道路区域，具有良好的实时性与可靠性.

Multilayer Perceptron Self-supervised Online Correction Road Recognition Algorithm

For autonomous mobile robot, the performances such as accuracy, robustness and real- time were crucial during the process of unstructured road recognition. A self-supervised online correction algorithm based on region of interest (ROI) and multi-layer perceptron (MLP) was proposed to solve the problem. Firstly, the effective computing region of processed image was defined by the ROI algorithm. Secondly, using multilayer perceptron to train the sample data to classify the region of interest according to the corresponding features, then using morphological processing and feature extraction to select the effective driving area. Lastly, the self- supervised online correction algorithm was used to correct the error processing results to ensure the accuracy of road classification and recognition. The experimental results show that the improved algorithm can accurately identify the road region in the environment. The algorithm is of high real-time and reliability