Robotic manipulators typically use continuous-range-of-motion actuators in combination with sophisticated control algorithms to track continuous trajectories with very high accuracy. This work discusses a very different approach to the design of robotic manipulators. There are many applications which do not require the tracking of continuous trajectories, e.g. simple inspection tasks or pick-and-place tasks in warehouses. This motivates the development of manipulators that use discrete actuators, i.e. actuators that are cheap, but only have a finite number of stable states. Such a manipulator is very robust and much cheaper than a traditional manipulator.; The first part of this thesis develops a framework for the use of discretely-actuated manipulators of the hybrid-serial-parallel type. Efficient methods are developed for workspace generation and inverse kinematics for this type of manipulator. These methods are also applicable to continuously-actuated manipulators. The second part addresses the design of spatial, discretely-actuated manipulators. This involves the selection of a suitable architecture and the development of a simulation tool to address a number of kinematic issues encountered in the design process. A spatial prototype consisting of six Gough (Stewart) mechanisms and actuated by 36 pneumatic cylinders is designed using this simulation tool.
展开▼
