Research On Humanoid Robot Path Planning Based On Embedded Vision

With the progress and development of science and technology, robot technologyhas been gradually incorporated into an important part of the field of socialdevelopment. And at the present stage, the field of humanoid robot has become the hotspot of the research which domestic and international technical personnel are fightingfor. Except for the advantages of its accurate,flexible, efficient and independent, theability of humanoid robot to adapt to the environment, complete the task is beyond thereach of human. Research on humanoid robot covers many fields, including machinelearning, artificial intelligence, sensor technology, communication technology, computergraphics, bionics and other related disciplines. With the development of human life,the demand has become increasingly stringent in all walks of life. The development ofhumanoid robot is facing more challenges now.Humanoid robot vision systems and humanoid robot plan planning are two majorresearch areas of the robot. Research of the paper which foucos on humanoid robot inthe embedded vision and path planning is as follows:(1) Chose the S3C2440demoboard. Set up hardware and software developmentplatform for embedded vision. Completed the cuts in the Linux kernel and file systemtransplantation and successfully transplanted OpenCV library and Qt in the demoboard.(2) In terms of robot vision, the research realized obstacle identification in front ofthe humanoid robot movement and ranging through the binocular vision algorithm inOpenCV. Introduction to the embedded system of binocular vision, the transplantationand the use of OpenCV library convenience to the inheritance and development ofalgorithm.Compared with the method of the traditional single vision chip, the cost hasbeen reduced and the application prospect has been expanded.(3) In terms of path planning of humanoid robot, aimed at the limitation of theperception to environment by humanoid robot binocular vision, the path planningalgorithm based on fuzzy control. The algorithm based on the experience of theenvironment and the obstacles has very strong adaptability. It can make the humanoidrobot achieve collision free motion in the process of movement.Proved by experiments, the method applied to humanoid robot can be very good tocomplete the task of the environment recognition and path planning, which furtherproves the validity and portability.