Gerard T. McKee
In this paper we describe research we have carried out in the area of human-robot interaction, specifically the provision of intelligence assisted viewing for human operators controlling remote robotics systems. We have developed intelligent support for camera placement based on a model, called visual acts, that treats the operator as an active interrogator of the remote environment. We describe a number of studies conducted to assess the model and an alternative, simpler model. Both models improve over manual control of the cameras by the operator, but show little relative difference in performance. Although the original motivation had been to provide intelligent robotic assistance to the operator, it is less clear now where the role of leadership did, could, or should lie. We suggest a flexible model whereby the locus of control shifts between human and the robot system based on the latter’s uncertainty over the former’ model of the task. The proposal also has implications for the locus of teaching and learning in human-robot systems.