Development of operator interfaces for a heavy maintenance manipulator

This dissertation details the development of an intuitive operator interface for a complex serial manipulator, to be used in heavy maintenance tasks. This interface allows the operator to control the manipulator in the 'task-space', with software handling the conversion to 'joint-space'. Testing of the interfaces shows operator task-space control to be most effective in reducing operator workload and improving the ease of use of a complex machine. These methods are applicable in concept, to a wider range of manipulators and other machines. A number of operator interfaces were developed: a Joystick Interface, a Master Arm interface and a 6-D Mouse Interface. The Joystick Interface made use of a task space to joint space transformation implemented in software. The Master Arm utilised a scale model to conduct the transformation. Finally, a 3D mouse Interface utilised sensors in an Android Device with a software based task to joint space transformation. These interfaces were tested and the Joystick Interface proved most suitable according to the operator's subjective opinion. Quantitative measurement also showed that it accurately reproduced the operator's commands. The software transformation developed for the Joystick and 6-D Mouse interfaces utilised the Jacobian Matrix to complete the task-space to joint-space conversion. However, since the manipulator contained a redundant joint, an additional algorithm was required to handle the redundancy. This additional algorithm also improved manipulator safety, as it navigated the arm away from singularities which could result in large joint movement. The novelty of this algorithm is based on its pragmatic approach, and could be modified to achieve a number of safety or performance goals. The control strategy centred on the operator specifying commands to the arm in the frame of the task. The developed algorithm enabled the control strategy by ensuring that viable solutions for joint velocity could be found in a manipulator that has redundant joints. Furthermore, this algorithm utilised a cost function that minimised the chances of large joint movements due to singularities, improving the safety of the device. Overall, the project has delivered a viable operator interface for controlling a complex, redundant manipulator. This interface was tested against a number of alternate operator interfaces. The contrasting results of the strengths and weaknesses of various interfaces meant that a number of key insights were gained, and a pragmatic approach to redundancy management was developed.