基于LVQ网络算法的目标分类方法

The Target Differentiation Method Based on LVQ Neural Network

随着机器人技术的不断发展,构建未知空间的环境地图以成为移动机器人技术中具有挑战性的研究课题。自主环境建模体现了机器人的感知能力和智能水平,在实际应用中具有十分重要的意义。文章研究了学习矢量量化算法在智能机器人对环境障碍物识别中的应用以及噪声、振幅、声纳TOF数据的偏移、距离以及目标体(锐角和边角)的角度等因素对系统分类性能的影响。实验表明,该方法对上述各种影响因素具备一定的鲁棒性,从而使得移动机器人能够在较大的距离范围内快速、可靠地识别室内各种典型障碍物。

With the development of science and technology,mapping-especially the target differentiation in unknown environments for mobile robots is a challenging subject in robotics. Mapping is embodies the perceive ability and the intelligent level of mobile robots. Thus,it is also play an important role in practical applications. In this paper,the robustness of the system to noise, partial removal of amplitude and TOF data and distance and different targets （edge with different,acute corner with different）has been studied. The experiment results show that the LVQ neural network has a high robustness to noise, and can differentiate the typical targets in a large ranging scope effectively and rapidly.