| This thesis presents the development and real-time implementation of a novel vision-based orientation sensor (VOS) for the absolute measurement of Atlas technology demonstration platform (Atlas TDP). The VOS takes the image of a set of binary code markers mounted on the surface of the Atlas TDP sphere. It then processes the image and obtains an estimation of the orientation of the sphere. A nonlinear quaternion-based kinematic controller is modified and implemented on the Atlas TDP using the orientation feedback from the VOS. In this controller, a gain scheduling scheme is developed to meet the requirements of fast response and small steady state error of the TDP system. In order to maintain the continuity of the orientation data flow in case the VOS misidentifies markers, a sensor fusion algorithm using unscented Kalman filter (UKF) is developed. This algorithm can also be used to generate more data samples between VOS measurements to match higher control system rates. Experiments were conducted to verify the performances of the VOS, the nonlinear quaternion-based kinematic controller, and the UKF sensor fusion algorithm. The details of the tests are given in this thesis, as well as the discussion of the results. |
