Femke E. van Beek

The effect of damping on the perception of hardness

In controlling teleoperation systems subject to communication delays, unstable behavior is often prevented by injecting damping. A proper perception of hardness is required to efficiently interact with an object, but it is unknown if and how this injected damping influences the perceived hardness of objects. To investigate the effect of damping on the perceived hardness of an object, 12 participants compared the hardnesses of lightly and heavily damped objects in a two-alternative forced choice experiment. Two tasks were investigated: an in-contact and a contact-transition task. For each task, 3 reference stiffnesses were used. Both force and position data were recorded. The results show that adding damping increases the perceived hardness of an object for an in-contact task, while it decreases perceived hardness for a contact-transition task. Movement and force data show that object indentation, mean and maximum velocity and adjusted rate-hardness, a new parameter defined in this paper, correlate with perceived object hardness. From a fundamental perspective, the results show that perceived hardness is influenced by damping in a task-specific way. Moreover, for designers of teleoperation systems the results provide insights for tuning and designing control architectures that properly reflect the desired hardness.

Haptic perception of force magnitude and its relation to postural arm dynamics in 3D

In a previous study, we found the perception of force magnitude to be anisotropic in the horizontal plane. In the current study, we investigated this anisotropy in three dimensional space. In addition, we tested our previous hypothesis that the perceptual anisotropy was directly related to anisotropies in arm dynamics. In experiment 1, static force magnitude perception was studied using a free magnitude estimation paradigm. This experiment revealed a significant and consistent anisotropy in force magnitude perception, with forces exerted perpendicular to the line between hand and shoulder being perceived as 50% larger than forces exerted along this line. In experiment 2, postural arm dynamics were measured using stochastic position perturbations exerted by a haptic device and quantified through system identification. By fitting a mass-damper-spring model to the data, the stiffness, damping and inertia parameters could be characterized in all the directions in which perception was also measured. These results show that none of the arm dynamics parameters were oriented either exactly perpendicular or parallel to the perceptual anisotropy. This means that endpoint stiffness, damping or inertia alone cannot explain the consistent anisotropy in force magnitude perception.

Adjusting Haptic Guidance to Idiosyncratic Visuo-Haptic Matching Errors Improves Perceptual Consistency in Reaching

When subjects reach for a visual target with their unseen hand, they make systematic errors (visuo-haptic matching errors). Visuo-haptic matching errors are idiosyncratic and consistent over time. Therefore, it might be useful to compensate for these subject-specific matching errors in the design of haptic guidance to make the guidance perceptually consistent with the visual information. In this study, we investigated whether compensating for visuo-haptic matching errors results in better perceptual consistency in a reaching task. Subjects (N = 12) had to reach for visual targets with the handle of a haptic device (PHANToM Premium 3.0/6DoF) held in their unseen dominant hand without guidance, with haptic guidance toward the target position, or with haptic guidance toward the position they would reach for according to their idiosyncratic visuo-haptic matching error. We found that the distance between the aiming point of the guidance and the reached end position was smaller for the guidance toward the idiosyncratic matched positions, suggesting a larger perceptual consistency. Adjusting for idiosyncratic visuo-haptic matching errors seems to have benefits over guidance to the visual target position.

Haptic Discrimination of Distance

While quite some research has focussed on the accuracy of haptic perception of distance, information on the precision of haptic perception of distance is still scarce, particularly regarding distances perceived by making arm movements. In this study, eight conditions were measured to answer four main questions, which are: what is the influence of reference distance, movement axis, perceptual mode (active or passive) and stimulus type on the precision of this kind of distance perception? A discrimination experiment was performed with twelve participants. The participants were presented with two distances, using either a haptic device or a real stimulus. Participants compared the distances by moving their hand from a start to an end position. They were then asked to judge which of the distances was the longer, from which the discrimination threshold was determined for each participant and condition. The precision was influenced by reference distance. No effect of movement axis was found. The precision was higher for active than for passive movements and it was a bit lower for real stimuli than for rendered stimuli, but it was not affected by adding cutaneous information. Overall, the Weber fraction for the active perception of a distance of 25 or 35 cm was about 11% for all cardinal axes. The recorded position data suggest that participants, in order to be able to judge which distance was the longer, tried to produce similar speed profiles in both movements. This knowledge could be useful in the design of haptic devices.

Correcting for Visuo-Haptic Biases in 3D Haptic Guidance

Visuo-haptic biases are observed when bringing your unseen hand to a visual target. The biases are different between, but consistent within participants. We investigated the usefulness of adjusting haptic guidance to these user-specific biases in aligning haptic and visual perception. By adjusting haptic guidance according to the biases, we aimed to reduce the conflict between the modalities. We first measured the biases using an adaptive procedure. Next, we measured performance in a pointing task using three conditions: 1) visual images that were adjusted to user-specific biases, without haptic guidance, 2) veridical visual images combined with haptic guidance, and 3) shifted visual images combined with haptic guidance. Adding haptic guidance increased precision. Combining haptic guidance with user-specific visual information yielded the highest accuracy and the lowest level of conflict with the guidance at the end point. These results show the potential of correcting for user-specific perceptual biases when designing haptic guidance.

The Effect of Global and Local Damping on the Perception of Hardness

In tele-operation systems, damping is often injected to guarantee system stability during contact with hard objects. In this study, we used psychophysical experiments to assess the effect of adding damping on the users perception of object hardness. In Experiments 1 and 2, combinations of stiffness and damping were tested to assess their effect on perceived hardness. In both experiments, two tasks were used: an in-contact task, starting at the objects surface, and a contact-transition task, including a free-air movement. In Experiment 3, the difference between inserting damping globally (equally throughout the workspace) and locally (inside the object only) was tested. In all experiments, the correlation between the participants perceptual decision and force and position data was also investigated. Experiments 1 and 2 show that when injecting damping globally, perceived hardness slightly increased for an in-contact task, while it decreased considerably for a contact-transition task. Experiment 3 shows that this effect was mainly due to inserting damping globally, since there was a large perceptual difference between inserting damping globally and locally. The force and position parameters suggest that participants used the same force profile during the two movements of one trial and assessed the systems reaction to this force to perceive hardness.

Subject-specific distortions in haptic perception of force direction

In a previous study, we found that the accuracy of human haptic perception of force direction is not very high. We also found an effect of physical force direction on the error subjects made, resulting in ‘error patterns’. In the current study, we assessed the between- and within-subject variation of these patterns. The within-subject variation was assessed by measuring the error patterns repeatedly over time for the same set of subjects. Many of these patterns were correlated, which indicates that they are fairly stable over time and thus subject-specific. The between-subject analysis, conversely, yielded hardly any significant correlations. We also measured general subject parameters that might explain this between-subject variation, but these parameters did not correlate with the error patterns. Concluding, we found that the error patterns of haptic perception of force direction are subject-specific and probably governed by an internal subject parameter that we did not yet discover.

Discrimination of Distance

While quite some research has focussed on the accuracy of haptic perception of distance, information on the precision of haptic perception of distance is still scarce, particularly regarding distances perceived by making arm movements. In this study, eight conditions were measured to answer four main questions, which are: what is the influence of reference distance, movement axis, perceptual mode (active or passive) and stimulus type on the precision of this kind of distance perception? A discrimination experiment was performed with 12 participants. The participants were presented with two distances, using either a haptic device or a real stimulus. Participants compared the distances by moving their hand from a start to an end position. They were then asked to judge which of the distances was the longer, from which the discrimination threshold was determined for each participant and condition. The precision was influenced by reference distance. No effect of movement axis was found. The precision was higher for active than for passive movements and it was a bit lower for real stimuli than for rendered stimuli, but it was not affected by adding cutaneous information. Overall, the Weber fraction for the active perception of a distance of 25 or 35 cm was about 11% for all cardinal axes. The recorded position data suggest that participants, in order to be able to judge which distance was the longer, tried to produce similar speed profiles in both movements. This knowledge could be useful in the design of haptic devices.

Visuo-Haptic Biases in Haptic Guidance

Visuo-haptic biases are observed when bringing your unseen hand to a visual target. The biases are different between, but consistent within participants. We investigated the usefulness of adjusting haptic guidance to these user-specific biases in aligning haptic and visual perception. By adjusting haptic guidance according to the biases, we aimed to reduce the conflict between the modalities. We first measured the biases using an adaptive procedure. Next, we measured performance in a pointing task using three conditions: (1) visual images that were adjusted to user-specific biases, without haptic guidance, (2) veridical visual images combined with haptic guidance, and (3) shifted visual images combined with haptic guidance. Adding haptic guidance increased precision. Combining haptic guidance with user-specific visual information yielded the highest accuracy and the lowest level of conflict with the guidance at the end point. These results show the potential of correcting for user-specific perceptual biases when designing haptic guidance.

Integrating Force and Position

In this study, we investigated the integration of force and position information in a task in which participants were asked to estimate the center of a weak force field. Two hypotheses, describing how participants solved this task, were tested: (1) by only using the position(s) where the force reaches the detection threshold, and (2) by extrapolating the force field based on perceived stiffness. Both hypotheses were also described formally, assuming a psychophysical function obeying a power law with an exponent smaller than one. The hypotheses were tested in two psychophysical experiments, in which 12 participants took part. In Experiment 1, an asymmetric force field was used and the presence of visual feedback about hand position was varied. In Experiment 2, a uni-lateral force field was used. For both experiments, hypothesis 1 predicts biases between (Experiment 1) or at (Experiment 2) the position(s) of the force detection threshold, while hypothesis 2 predicts smaller biases. The measured data show significant biases in both experiments that coincide with the biases predicted by using force detection thresholds from literature. The average measured responses and their variabilities also fitted very well with the mathematical model of hypothesis 1. These results underline the validity of hypothesis 1. So, participants did not use a percept of the stiffness of the force field, but based their estimation of the center of the force field on the position(s) where the force reached the detection threshold. This shows that force and position information were not integrated in this task.