Research On Self-Localization Of Service Robot Based On Kinect Depth Vision

With the emergence of a global population ageing, in order to improve the quality of old man's life, demand of service robots is increasing. At present, it's paid attention that study and develop the service robots facing to the family and the repetitive family work or "coolies work" shall be borne by the service robot more and more. Service robot can only finish the special mission successful after make the instantaneous position clear and know how to move from instantaneous location to target location. So that, self-localization is the basis of the action how the service robot finish the route planning, self-guide and so on. The solution is the premise that mobile robot finish the mission.Therefore, in this paper, self-localization of service robot in indoor structured environments are studied, and builds an indoor orientation system which base on Kinect depth visual sensor and manually activated beacon. put forward and improve the trilateral positioning algorithm. the algorithms improve autonomous location ability of the robot in orientation environments. The main contents are as follows:Firstly, this paper expound the service robots' current situation of study and use domestically and abroad. Analyze and study key technology of self- positioning for the service robot. And expound the significance self-positioning for the service robot.Secondly, this paper expound the service robot hardware structure, model of coordinate system, dynamatic model, kinect visual sensor ranging technology, operating principle. Use kinect sensor for the ranging experiment.Thirdly, in order to solve the problem that the low accuracy and the limited scope of the traditional edge detection algorithm. Put forward a improved method based on Sobel algorithms. Use the improved method to do an experiment on photos which are achieved by Kinect sensor. The results show that the algorithm is effective.Fourthly, the path planning techniques of service robot are studied in the related theoretical knowledge and calculation methods of image registration. the path planning techniques of mobile robot are studied in the related theoretical knowledge. Point out the advantage and disadvantage of two methods. Put forward the method of combining with the feature of edges and image registration method of mutual information. Verify the effectiveness and accuracy of image registration by improving algorithm.At last, study beacon localization and vision localization in depth. Introduce self-localization system of this paper in detail, builds an indoor orientation system which base on Kinect depth visual sensor and manually activated beacon. And introduce three localization method based on ranging localization method. Improve the problem usually happen in the trilateral localization algorithm. Verify the feasibility of localization by matlab simulation experiment.