Research On Autonomous Walking Of Humanoid Robot Based On Vision-SLAM

With the development of the science and technology, national and international scholars have more and more intention on the research of the intelligent robots. Robot with human external characteristics has been developed rapidly. And many people love those robots. Among them, the research of human robot autonomous walking has become a hot spot in the research field of human robot. The robot vision research has great significance on the research of human robot autonomous walking.This paper use EKF-SLAM algorithm to realize the simultaneous localization and mapping in the indoor environment. This paper recognizes the different landmark respectively using Color-contour method and Harris-SIFT algorithm. The robot obtains the landmarks location and the environment map information through the laser sensor on the robot head. Then robot can create the environment map using the iterative segmentation method and the least square method. Finally, the experiments using the NAO robot verify the effectiveness of the proposed method in this paper. The specific research contents are as follows:Firstly, the NAO robot recognizes the object of single color and regular shape in the environment based on the Color-contour method. The NAO robot obtains the location information of the target object using the laser sensor. And the Color-contour method and the laser sensor are combined with the EKF-SLAM algorithm to realize the autonomous walking of the NAO robot. We set the landmark and disruptor in the experiment environment to realize the proposed method. The results show that the NAO robot only recognizes the landmark. The NAO robot will use laser sensor to obtain the location information of landmarks. Then the NAO robot will realize the autonomous walking. This method not only can effectively recognize the object, but also reduces the amount of computation.Secondly, the NAO robot build the local environment map based on environment information obtained by the laser sensor. We establish the URG-04 LX laser sensor model. The laser sensor model can realize the transformation of the laser coordinate system, the NAO robot coordinate system and the world coordinate system. We use laser sensor model to process the data information of angle, distance and coordinates. The data is acquired by the laser sensor. Then we use the iterative segmentation method and the least square method to build the environment map. Finally, we verify the proposed method by the experiment.Finally, we propose the Harris-SIFT algorithm to improve the recognition range of human robot in the NAO robot autonomous walking. The Harris-SIFT algorithm can recognize the object of irregular shape and complex image. The combination of the Harris-SIFT algorithm and the laser sensor can avoid all objects information obtained in the environment. The NAO robot can realize to obtain the location information of the specific object in the indoor environment. The Harris-SIFT algorithm can improve the intelligence of the obtained information and anti-disturbance ability. Then we combine the Harris-SIFT algorithm with the EKF-SLAM algorithm to complete the autonomous walking of the NAO robot in the indoor environment.