Implementati On Of A Dual Camera Coordination Of Intelligent Robot Hand Grasping Method

With the continuous improvement of the hardware configuration of the robot, the robot existing software technology can not meet the expectations of the ability of a robot man. People expect robots to be more autonomous and more intelligent to complete tasks, rather than relying on the set steps to complete the assigned task. One of the most basic ability crawling robot to perform other tasks as necessary, the level of their ability is an important basis for evaluation of robotic intelligence; Meanwhile, as the robot vision sensor outside the main channel to obtain information, the information acquisition and processing capacity has also become an important symbol of robot intelligence; therefore how to enhance the capacity of the visual sensor-based robot crawling objects have theoretical significance and practical significance.Most of the current research and application of robots used only with a single camera, grab positioning technology based on a single camera purpose, the main idea is to combine the camera or object motion model and multi-frame image information, obtaining three dimensional information about the target; it needs camera calibration and motion modeling and image fusion technology involved, the positioning process large amount of computation, and susceptible to outside influence. However, if the use of fuzzy positioning and objectives ranging from monocular camera horizontal and vertical angle of vision center combines information may not involve in the case of single-purpose positioning multi-frame camera calibration and image fusion; this will reduce monocular positioning requirement has universal application.Current crawling robot intelligent control case study very little, especially in the movement of the target and the robot case studies. Therefore, Based on the humanoid robot in-depth study of each neighborhood intelligent control method, the mainstream technology roadmap and technical characteristics, etc. Try fuzzy neural network control algorithm is introduced to crawl control system, achieve "watch and do" simulate human thinking smart grab a moving target function withnot the camera calibration, improve the control system reusability, reduce the amount of computation algorithms to improve the real-time systems.Research and testing platform is based on the NAO humanoid robot with the embedded Gentoo GNU/Linux systems. First, in-depth analysis NAO robot visual servo system hardware capabilities and limitations, to build a new auxiliary field for NAO robot and solve the NAO robot is not open-source operating system and not supported peripherals and other problems. Secondly, combining the advantages of dual robotics humanoid robot using both hands to grab coordinated manner in the control algorithm developed on the basis of further improve the crawl rate and improve the utilization of execution at the end of the robot, humanoid robot hardware in line with the designer’s mind. Finally, the goal of making search, locate, track and capture methods in the NAO robot platform experiment data show the feasibility and effectiveness of the control scheme, the superiority and universality, can apply this method to other robotic systems with embedded dual manipulator.