The Effect of Trial-by-Trial Adaptation on Conflicts in Haptic Shared Control for Free-Air Teleoperation Tasks.

Haptic shared control can improve execution of teleoperation and driving tasks. However, shared control designs may suffer from conflicts between individual human operators and constant haptic assistance when their desired trajectories differ, leading to momentarily increased forces, discomfort, or even deteriorated performance. This study investigates ways to reduce conflicts between individual human operators and a haptic shared controller by modifying supported trajectories. Subjects (n=12) performed a repetitive movement task in an abstract environment with varying spatio-temporal constraints, both during manual control and while supported by haptic shared control. Four types of haptic shared control were compared, combining two design properties: the initial supported trajectory (either the centerline of the environment or an individualized trajectory based on manual control trials), and trial-by-trial adaptation of guidance towards previously performed trajectories (either present or absent). Trial-by-trial adaptation of guidance reduced conflicts compared to non-adaptive guidance, whether the initial trajectory was individualized or not. Without trial-by-trial adaptation, individualized trajectories also reduced conflicts, but not completely: when guided, operators adapt their preferred trajectories. In conclusion, trial-by-trial adaptation is the most promising approach to mitigate conflicts during repetitive motion tasks.

Haptic Shared Control in Tele-Manipulation: Effects of Inaccuracies in Guidance on Task Execution.

Haptic shared control is a promising approach to improve tele-manipulated task execution, by making safe and effective control actions tangible through guidance forces. In current research, these guidance forces are most often generated based on pre-generated, errorless models of the remote environment. Hence such guidance forces are exempt from the inaccuracies that can be expected in practical implementations. The goal of this research is to quantify the extent to which task execution is degraded by inaccuracies in the model on which haptic guidance forces are based. In a human-in-the-loop experiment, subjects (n = 14) performed a realistic tele-manipulated assembly task in a virtual environment. Operators were provided with various levels of haptic guidance, namely no haptic guidance (conventional tele-manipulation), haptic guidance without inaccuracies, and haptic guidance with translational inaccuracies (one large inaccuracy, in the order of magnitude of the task, and a second smaller inaccuracy). The quality of natural haptic feedback (i.e., haptic transparency) was varied between high and low to identify the operator's ability to detect and cope with inaccuracies in haptic guidance. The results indicate that haptic guidance is beneficial for task execution when no inaccuracies are present in the guidance. When inaccuracies are present, this may degrade task execution, depending on the magnitude and the direction of the inaccuracy. The effect of inaccuracies on overall task performance is dominated by effects found for the Constrained Translational Movement, due to its potential for jamming. No evidence was found that a higher quality of haptic transparency helps operators to detect and cope with inaccuracies in the haptic guidance.

A task-specific analysis of the benefit of haptic shared control during telemanipulation.

Telemanipulation allows human to perform operations in a remote environment, but performance and required time of tasks is negatively influenced when (haptic) feedback is limited. Improvement of transparency (reflected forces) is an important focus in literature, but despite significant progress, it is still imperfect, with many unresolved issues. An alternative approach to improve teleoperated tasks is presented in this study: Offering haptic shared control in which the operator is assisted by guiding forces applied at the master device. It is hypothesized that continuous intuitive interaction between operator and support system will improve required time and accuracy with less control effort, even for imperfect transparency. An experimental study was performed in a hard-contact task environment. The subjects were aided by the designed shared control to perform a simple bolt-spanner task using a planar three degree of freedom (DOF) teleoperator. Haptic shared control was compared to normal operation for three levels of transparency. The experimental results showed that haptic shared control improves task performance, control effort and operator cognitive workload for the overall bolt-spanner task, for all three transparency levels. Analyses per subtask showed that free air movement (FAM) benefits most from shared control in terms of time performance, and also shows improved accuracy.