In contrast to automated production, human intelligence is deemed necessary for successful execution of assembly tasks that are difficult or expensive to automate in small and medium lots. However, human ability is hindered in some cases by physical barriers such as miniaturization or in contrast, very heavy components. Telepresence technology can be considered a solution for performing a wide variety of assembly tasks where human intelligence and haptic sense are needed. This work highlights several issues involved in deploying industrial telepresence systems to manipulate and assemble microparts as well as heavy objects. Two sets of experiments are conducted to investigate telepresence related aspects in an industrial setting. The first experiment evaluates the usefulness of haptic feedback for a human operator in a standard pick-and-place task. Three operation modes were considered: visual feedback, force feedback, and force assistance (realized as vibration). In the second experiment, two different guidance strategies for the teleoperator were tested. The comparison between a position and a velocity scheme in terms of task completion time and subjective preferences is presented.

In telepresence and teleaction (TPTA) systems, the transmission of haptic signals puts high demands on the applied signal processing and communication procedures. When running a TPTA session across a packet-based communication network (e.g., the Internet), minimizing the end-to-end delay results in packet rates of up to the applied sampling rate of the local control loops at the human system interface and the teleoperator. The perceptual deadband data reduction approach for haptic signals successfully addresses the challenge of high packet rates in networked TPTA systems and satisfies the strict delay constraints. In this paper, we extend the underlying perceptual model of the deadband approach by incorporating psychophysical findings on human force-feedback discrimination during operators' relative hand movements. By applying velocity-dependent perception thresholds to the deadband approach, we observe further improvement in efficiency and performance due to improved adaption to human haptic perception thresholds. The psychophysical experiments conducted reveal improved data reduction performance of our proposed haptic perceptual coding scheme without impairing the user experience. Our results show a high data reduction ability of up to 96% without affecting system transparency or the operator's task performance.