Position Detection System Based On Vision Of Parallel Mechanism

Position is one of important parameters that reflect the state of parallel robot. It isimportant for calibration of the robot and visual servo, and its control accuracy is one ofkey specifications of measuring robotics performances. Therefore, it is important tostudy the position detection method of parallel robot platform. But generally it isdifficult to obtain all the positional parameters of parallel mechanism. The vision-basedposition measurement technology can get all the positional parameters of the platform,and has the advantages of real-time, non-contact measurement and high degree ofautomation. It is ideally for the position detection of parallel mechanism.In this thesis, based on the analysis of motion characteristics of3-PRS parallelmechanisms, a visual position measurement system on parallel mechanism is designedand a kind of monocular vision measuring model is established. This model with simplestructure does not require stereo matching and can satisfy the need of positionmeasurement system. Quaternion is introduced into the rotation matrix to calculate therelative pose matrix of the robot, and then the closed-form solutions for the position andorientation of the target are obtained. The target and efficient feature points matchingmethod are designed. Image processing algorithms are researched for the requirementsof visual system. An embedded vision hardware system based on DSP is designed in thethesis. Embedded software codes are also given. A simple and reliable vision system isbuilt and a user-friendly software interface is developed. Experiments are performed onthe XY servo platform and turning platform to verify the correctness of the algorithm.Finally experiments are carried out on3-PRS parallel mechanism. The accuracy of theagency and error sources are analyzed.