Research On Indoor Robot Vision Navigation And Positioning Based On Embedded System

Robot technology is widely used in high risk environment such as poisonous or anoxic rooms to replace human being to finish tasks continuously, stably and efficiently. Navigation and positioning based on sensor technology are two major challenges for robot to find certain target accurately and ensure completion of task. As to all kinds of robot sensory, a significant research area---robot vision owns incomparable advantages of large range measurement and vast quantities of information to recognize the environment which equal important. The hardware and software of embedded systems are low cost, small in size, convenient interconnection and highly integration and usually designed for specific applications. Based on these reasons, it is an appropriate choice for robots’control. Robot system with embedded technology is becoming the mainstream robot research field. While there is little paper concerning about this research because of the difficulties in robot vision. This paper studies on indoor robot vision navigation and positioning and is organized with two technical aspects: embedded system and robot vision.Firstly, the paper introduces the design of embedded system for indoor robot containing transplanting bootloader, linux kernel and root filesystem. Base on the embedded hardware system, cameras and motor drive equipment program are developed. Also, OpenCV image processing library, WIFI and other services (e.g. sshd) necessary for robot system are transplanted on it.Secondly, based on the embedded robot system, the paper conducts camera calibration design, robot vision image feature extraction algorithm design and detailed analysis of robot vision white line path tracking problem. After the analysis of indoor environmental features, improved hough white line extraction algorithm is proposed based on statistical methods. Experiment shows that the proposed hough white line extraction algorithm processing ability is weak. We finally optimize the algorithm to reduce image processing dimension and realize strong feasibility on condition of correct white string extraction. With the white line position information, the robot vision system is calibrated and robot position and posture in the world coordinate system are obtained.Finally, the system setup and software development method are described. Meanwhile, robot remote operation platform is developed. We conduct navigation and positioning experiment to analyze the system performance. The results indicate the validation of the proposed algorithm. Research on indoor robot visual navigation and positioning based of embedded system can be used in office, home, entertainment place, museum and etc.