Study On Monocular Vision For Family Service Robot Navigation

With the current trend of population aging and the gradual improvement of the quality of life requirements,demand for home service robot is becoming increasingly urgent.The role robots plays in family is not confined to simple clean job such as sweeping the floor,washing the dishes,even provide entertainment for people,create a more safe and comfortable environment.Whatever kind of service home service robot eventually give people,navigation technology is the precondition and foundation of the whole process,which is also key technology to realize real intelligence for home service robot.Therefore,the research on navigation for home service robot has great market potential and important research value.This paper studies the navigation problem of family service robot with single vision.It mainly includes the following aspects:1.The research background and significance of this paper is elaborated.The national research status of home service robot and visual navigation is introduced.The necessity and main research contents of this paper is explained.2.The home service robot based on monocular vision is designed in detail.The design and optimization of the robot's mechanical mechanism is designed.The hardware in the system model is selected,while the important relevant parameters are given.The software architecture and image processing is optimized.Humanized upper computer interface is designed.3.Modeling and analysis of the family service robot is down in this paper,including the optimal layout and kinematics modeling analysis of the platform,subsequently D_H modeling for the mechanical arm,and eventually introducing imaging model of the camera,in which the binomial fitting method is used to calibrate the monocular camera of the system.4.The three main problems in the navigation of family service robot are the locationand map reconstruction,obstacle avoidance,as well as path planning.For localization and map reconstruction,the method of integration of inertia is adopted combined with monocular vision.By getting use of optical flow,the obstacle detection is carried out,then the balance strategy is applied to avoid the obstacle,finally the autonomous obstacle avoidance function of the robot can be completed.Aimed at the environment with complex obstacles,a path manage method based on Mean Shift is proposed,and the specific steps of the algorithm are described in detail.5.The autonomous navigation and obstacle avoidance performance of the robot was tested,and the test process was analyzed in detail.Full coverage of service robot traversal path planning and the shortest collision free path planning is examed by the simulation experiments.In order to make the experiment more intuitive,visual programming in MATLAB environment is adopted.The results of the experimental indicated that the improved algorithm can efficiently realize navigation task for robot.