An Indoor Localization Method for Humanoid Robot Based on Artificial Landmark

Abstract

In the paper, we try to reduce errors of localization of humanoid robot. As we know, there are many errors when a humanoid robot trying to locate itself indoor, this paper presents a method of indoor localization for humanoid robot based on some artificial landmarks in local region. We improve the accuracy of localization by deploying landmarks in some local key areas, or placing them in some key positions, and extract feature information of some special objects as natural landmarks. When a humanoid robot trying to determine its position in indoor environment, using the local landmarks can reduce the cumulative error, and improve the accuracy of localization. Building feature map using current popular feature points method, and using SURF to improve the speed, applying the particle filter method to complete global localization. The results show that the method is effective, it can improve t positioning accuracy.