Shared autonomy architecture for skill execution manipulator

In this paper, we report a shared autonomy architecture for skill execution manipulator. A skill transfer method has been proposed in our previous research. Many skills for a plant maintenance also have been developed. Our goal is that an autonomous robot implemented the skills executes maintenance tasks in the plant. When the robot fails the task execution, the robot should recognize all conditions and recover by itself. However, in current technology, the robot to recover from all failed conditions could not be developed. Then, an idea of shared autonomy architecture for our research area is introduced. In this idea, when the robot fails to perform the task and can not recover from failed conditions, an operator intervenes to control the manipulator. The operator moves the manipulator to a convenient condition in a bilateral teleoperation. After the operator´s motion, the robot re-executes the target task. To demonstrate our proposed architecture, we choose a nut attachment task.