Search

Data-driven texture modeling and rendering has pushed the limit of realism in haptics. However, the lack of haptic texture databases, difficulties of model interpolation and expansion, and the complexity of real textures prevent data-driven methods from capturing a large variety of textures and from customizing models to suit specific output hardware or user needs. This work proposes an interactive texture generation and search framework driven by user input. We design a GAN-based texture model generator, which can create a wide range of texture models using Auto-Regressive processes. Our interactive texture search method, which we call “preference-driven,” follows an evolutionary strategy given guidance from user's preferred feedback within a set of generated texture models. We implemented this framework on a 3D haptic device and conducted a two-phase user study to evaluate the efficiency and accuracy of our method for previously unmodeled textures. The results showed that by comparing the feel of real and generated virtual textures, users can follow an evolutionary process to efficiently find a virtual texture model that matched or exceeded the realism of a data-driven model. Furthermore, for 4 out of 5 real textures, $\geq$ 80% of the preference-driven models from participants were rated comparable to the data-driven models. [ABSTRACT FROM AUTHOR]

Kinesthetic haptic devices are designed primarily to display quasistatic and low-bandwidth forces and moments. Existing methods for vibrotactile display sometimes introduce haptic and/or audio artifacts. In this article, we propose a method to display vibrotactile stimulus signals of moderate to high frequency (20–500 Hz) using kinesthetic haptic devices with a standard 1 kHz haptic update rate. Our method combines symmetric square-wave signals whose periods are even multiples of the haptic update period with asymmetric square-wave signals whose periods are odd multiples of the haptic update period, while ensuring that the positive and negative impulses are balanced in both cases, and utilizing the just noticeable difference in frequency discrimination to avoid the need to display other frequencies. For frequencies at which the above method is insufficient, corresponding to a small band near 400 Hz for a 1 kHz update rate, we utilize a signal-mixing method. Our complete method is then extended to render haptic gratings by measuring scanning velocity, converting the local spatial frequency to its equivalent instantaneous temporal frequency, and displaying a single full-period vibration event. In a series of human-subject studies, we showed that our proposed method is preferred over existing methods for vibrotactile display of signals with relatively high-frequency content. [ABSTRACT FROM AUTHOR]

In our recent article, we found a general quadratic weighting function to predict how a six-dimensional (6D) vibrotactile stimulus rendered at the haptic interaction point (HIP) of a kinesthetic haptic interface, with the stylus held in a precision pen-hold grasp, is mapped to an equivalent one-dimensional (1D) stimulus that is normalized by the detection threshold. However, in that work we did not constrain the weighting function to be positive semidefinite, and as a result, the model will not generalize well to all future 6D inputs. In this addendum, we reconsider the original data set, incorporating the positive-semidefinite constraint. We find that as few as four independent parameters are required to describe the coupling, and the model has one fewer coupling term than originally proposed. We also describe the process of fitting a positive-semidefinite function to a new stylus at some specific excitation frequency. [ABSTRACT FROM AUTHOR]

The human hand detects high-frequency vibrations in all directions but cannot distinguish the direction, which suggests a multi-dimensional vibrotactile stimulus is haptically equivalent to some one-dimensional (1D) stimulus. In this article, we explore how a 6D vibrotactile stimulus rendered at the haptic interaction point (HIP) of a kinesthetic haptic interface, with the stylus held in a precision pen-hold grasp, is mapped to an equivalent 1D stimulus normalized by the detection threshold. We gather a large human-subjects data set in which we determine detection thresholds for 45 distinct combinations of three orthogonal forces and three orthogonal torques rendered at the HIP, at a single frequency of 108 Hz corresponding to the peak sensitivity in our prior study. Using this data set, we find a general quadratic weighting function to predict the 1D normalized stimulus for a given 6D vibrotactile stimulus. We find that including just seven (out of a possible 21) independent parameters in the symmetric weighting matrix is sufficient to capture the non-obvious coupling between forces and torques rendered at the HIP for dimensional reduction from 6D to 1D. [ABSTRACT FROM AUTHOR]

Interacting with physical objects through a tool elicits tactile and kinesthetic sensations that comprise your haptic impression of the object. These cues, however, are largely missing from interactions with virtual objects, yielding an unrealistic user experience. This article evaluates the realism of virtual surfaces rendered using haptic models constructed from data recorded during interactions with real surfaces. The models include three components: surface friction, tapping transients, and texture vibrations. We render the virtual surfaces on a SensAble Phantom Omni haptic interface augmented with a Tactile Labs Haptuator for vibration output. We conducted a human-subject study to assess the realism of these virtual surfaces and the importance of the three model components. Following a perceptual discrepancy paradigm, subjects compared each of 15 real surfaces to a full rendering of the same surface plus versions missing each model component. The realism improvement achieved by including friction, tapping, or texture in the rendering was found to directly relate to the intensity of the surface's property in that domain (slipperiness, hardness, or roughness). A subsequent analysis of forces and vibrations measured during interactions with virtual surfaces indicated that the Omni's inherent mechanical properties corrupted the user's haptic experience, decreasing realism of the virtual surface. [ABSTRACT FROM PUBLISHER]

Modern haptic interfaces are adept at conveying the large-scale shape of virtual objects, but they often provide unrealistic or no feedback when it comes to the microscopic details of surface texture. Direct texture-rendering challenges the state of the art in haptics because it requires a finely detailed model of the surface's properties, real-time dynamic simulation of complex interactions, and high-bandwidth haptic output to enable the user to feel the resulting contacts. This paper presents a new, fully realized solution for creating realistic virtual textures. Our system employs a sensorized handheld tool to capture the feel of a given texture, recording three-dimensional tool acceleration, tool position, and contact force over time. We reduce the three-dimensional acceleration signals to a perceptually equivalent one-dimensional signal, and then we use linear predictive coding to distill this raw haptic information into a database of frequency-domain texture models. Finally, we render these texture models in real time on a Wacom tablet using a stylus augmented with small voice coil actuators. The resulting virtual textures provide a compelling simulation of contact with the real surfaces, which we verify through a human subject study. [ABSTRACT FROM PUBLISHER]