Study Of Multifunctional Robot Platform And It’s Key Technologies

Nowadays, the mobile robot is developing towards an integrated direction of intelligent, multifunctional, collaborative and networked. With the strong environmental adaptability, varieties of special tasks can be completed by the multifunctional intelligent mobile robot in complex environment. So that it becames to an important development branch and a hot spot of robot researching. A four-wheel-based mobile robot platform has been developed in this paper. The depth detection and3D reconstruction of a space objects during the moving of the robot has been achieved via the binocular stereo vision technology. A C/S-based communication system has been built in a wireless LAN, which using of asynchronous non-blocking sockets. The general design of the mobile robot’s platform is introduced, and it’s key technologies has been studied in this paper, emphatically discusses how to realize the distance measuring based on binocular stereo and high-speed image data compression and communications.Wheeled structure is used to improve the moving speed and flexibility of the mobile robot. STC12C5A60S2MCU is used as the control core to control moving and turning by measure of pulse width modulation. The USB COMS binocular cameras is used as image acquisition unit to reduce the power consumption, the processing of image acquisition and data transmission are assigned to different threads by the multithreading technology so as to avoid the blocking in main thread.For the purpose of ensuring the accuracy of vision measurement, the camera model should be built to calibrate the camera, and correct image data by calibration parameters. For solving the problem of slow speed base on the traditional calibration and correction algorithm, a GPU-based parallel image correction method is put forward in this paper. The distance measuring principle of binocular stereo vision is studied, stereo rectification and the ideal binocular measurement model is used to realize the distance measure and space point reconstruction, so that it can satisfy the near real-time requirements in the course of the campaign.The data of color image is very huge brings inconvenience to store or transfer. So it’s necessary to encode the image data for eliminating the redundant information, so as to reduce the burden of the network. The static image compression algorithm JPEG is recommended, and a parallel JPEG algorithm based on CUDA is designed to speed up the JPEG compression algorithm. In addition, the mainstream video compression H.264is introduced and used to design a video compression program which can realize the data compressing of the video base on x264. In the network communications, different transmission schemes are designed for different characteristics of receivers. For remote monitoring, the image data is rapidly transfer by UDP. To ensure the data restructured correctly by receivers, a special packet header is designed to the front of data package. The information of data subcontracting order is wrote to the special packet header by the sender, and the data packet can be restructured via the data subcontracting order in the special packet header when the receiver gets the data. TCP is used to transfer the important files and commands or the image used to analysis-aid, because it can protect the right arriving order of the data. A TCP-based parallel data transmission is designed to increasing the transmission speed by usimg of the multi-threading technology.