Visual Identification Research And Design Of The GD-1Humanoid Robot

From the mid-eighties of the last century, the international community has been setting off a wave of researching the intelligent robot, paying more attention to the Humanoid Robot. Humanoid robot is an important branch of the robot, the highest state of robotics research. It is a concentrated expression of the theory and technology of researching intelligent robots. It will bring the development and progress of many related disciplines and technologies. The head of the Humanoid Robot is an important component of the Humanoid Robot, looking at it from the appearance, which is the most intuitive sign of personification degree. It will give people a feeling of intimacy and will eliminate the psychological barriers between human and robots, which will help improve their social skills. The human vision plays an important role in guiding human behavior, and the robot vision system will give the robot a high level of sensory functions that allow the robot by the intelligent and flexible way to respond to its surroundings.The silicone robot head, in our laboratory has been of the function of eye movements, but no the vision function. The main target of this research project is to add the vision function for the silicone robot in our laboratory. The robot with vision will accomplish the related action or task based on the instructions of the obtained image information, so that the Humanoid Robot will be of a higher level of intelligence.The development of the Humanoid Robot vision system needs digital image processing and pattern recognition technologies. This paper mainly research image processing algorithms and text recognition method related to computer vision. After a series of image pre-processing, we will get normalized single character image information. It will be as the input of the character recognition system. This system mainly identifies some direction character information. According to the characteristics of this system, we use artificial neural network classifier to identify the direction characters information. The encoded recognition results transmit to the robot control layer to guide the robot action, making the existing laboratory humanoid robots more intelligent.