Abstract
This dissertation concerns the critical topics of a Stewart platform-type manipulator: the for-ward kinematic analysis and self-calibration strategies. The thesis begins with a broad overview of existing researches regarding a Stewart platform, where important issues are discussed from different perspectives. All the relevant discussions will be helpful to provide a clear insight into this kind of parallel manipulator system. The forward kinematic analysis of a Stewart platform can have up to 40 solutions, which can not be expressed in an explicit form. Consequently, the goal of the related studies is to seek a solution scheme with high efficiency and accuracy. In the present thesis, three auxiliary sen-sors are introduced for this purpose to achieve the following merits: 1) the solution scheme is suitable for any general fully parallel manipulator; 2) it leads to a unique and closed-form solu-tion with remarkable efficiency and accuracy; and 3) the optimum location of the redundant sensor is also suggested, which makes the approach insensitive to the misalignment of the sen-sor location and the measurement errors. Accordingly, the method is suitable for applications where real-time computation is demanded. The notion of self-calibration is particularly suitable for a Stewart platform due to its in-herent closed-loop kinematic chains. However, certain significant limitations exist in the previous literatures. This thesis first thoroughly reveals these drawbacks, thereby proposes two novel strategies to overcome such limitations. It turns out that, in a workspace of five de-grees-of-freedom, the proposed schemes are cost-effective and attractive for an autonomous hexapod manipulator when great precision is required. Since an efficient approach for the forward kinematic analysis and accurate parameters of the system are both the prerequisites of designing an excellent controller, the investigations of the thesis are vitally important to a Stewart platform that serves as a precision machine.