Research On Humanoid Robot Motion Planning Based On Vision

With the rapid development of information technology, the robot will become an important area of social development component. And at this stage of many domestic and foreign researchers are also engaged in the field of robotics research. Humanoid robotics researchers been able to become a research focus, in addition to its own motion flexible, efficient autonomous advantages, but also very strong ability to complete the task. Humanoid robot research includes many fields, including machine learning, robotics, artificial intelligence, sensor technology, computer vision technology, multi-agent systems, image recognition. Computer hardware and software development in the era of technology continues to improve, so that people increasingly high demand for the robot, but also the humanoid robot research workers face greater challenges.Humanoid robot vision systems and humanoid robot motion planning are two major research areas of the robot in real-life applications have a significant role. This article is based HIT-Ⅱ humanoid robot, humanoid robot in the visual image processing and motion planning aspects of the following areas of research and improvements:(1) We will HIT-Ⅱ humanoid machine comes with the camera as a visual, visual humanoid robot used in the machine vision system is more suitable HSI color space. In the image pre-processing, is used for image denoising based on fractional integral approach to the de-noising, compared with traditional methods, this method can increase the SNR, retaining the original image Information about the edge and texture.(2) performing feature extraction and matching process, this paper is based on the scale invariant feature transform SIFT operator, and the operator of the traditional SIFT descriptor in the generation process has been optimized to increase its speed. For traditional SIFT point matching algorithm efficiency is low, slow speed matching problems, proposed and implemented the SIFT-based regional growth point matching algorithm. This algorithm can match the reliability of the basic premise remains the same, significantly improve matching speed.(3) In the humanoid robot motion planning for humanoid robot sensors perceive the limitations of the obstacle problem, we propose a method based on the information window. This algorithm improved humanoid robot for obstacle sensing algorithm that humanoid robot in motion, do not touch the greatest degree of movement.Experiments show that through the application of the above method in HIT-Ⅱ humanoid robot, the robot performance has been significantly improved. And by humanoid robot obstacle avoidance experiment with football has been further verified.