Effect of Avatar Appearance on Detection Thresholds for Remapped Hand Movements
Effect of Avatar Appearance on Detection Thresholds for Remapped Hand Movements
复制标题
头像外观对重新映射的手部动作检测阈值的影响
DOI：
10.1109/tvcg.2020.2964758
复制
发表时间：
2020
影响因子：
5.2
通讯作者：
Michitaka Hirose
中科院分区：
文献类型：
--
作者：
Nami Ogawa;Takuji Narumi;Michitaka Hirose
虚拟现实中的手部交互技术通常利用视觉对本体感觉的优势来将物理手部运动重新映射到不同的虚拟运动。然而，当虚拟手和物理手之间的偏移增加时，重新映射的虚拟手运动很难自归因，并且用户变得意识到重新映射。有趣的是，身体的自我归属感在心理学领域被称为身体所有权感（SoBO），并且化身越逼真，SoBO越强。因此，我们假设现实的化身（即，人手）可以比抽象化身（即，球形指针），从而使得重新映射不太明显。在这篇文章中，我们提出了一个实验中，参与者反复执行达到运动与他们的右手，而不同数量的水平转移。我们测量了移动方向（左或右）和化身外观（现实或抽象）的每个组合的重映射检测阈值。结果表明，逼真的化身提高了检测阈值（即，降低了灵敏度）比应用了平移时的抽象化身降低了31.3%（即，当手沿远离身体中线的方向移动时）。此外，本体感受漂移（即，朝向化身的自定位的位移）对于真实化身的自定位位移而言更大，这表明在真实化身中的视觉本体感受整合期间视觉信息被给予更大的偏好。我们的研究结果定量地表明，逼真的化身可以使重新映射不太明显的虚拟和物理运动之间的较大的不匹配，并可以潜在地改善各种各样的手动重新映射技术，而不改变映射本身。
Hand interaction techniques in virtual reality often exploit visual dominance over proprioception to remap physical hand movements onto different virtual movements. However, when the offset between virtual and physical hands increases, the remapped virtual hand movements are hardly self-attributed, and the users become aware of the remapping. Interestingly, the sense of self-attribution of a body is called the sense of body ownership (SoBO) in the field of psychology, and the realistic the avatar, the stronger is the SoBO. Hence, we hypothesized that realistic avatars (i.e., human hands) can foster self-attribution of the remapped movements better than abstract avatars (i.e., spherical pointers), thus making the remapping less noticeable. In this article, we present an experiment in which participants repeatedly executed reaching movements with their right hand while different amounts of horizontal shifts were applied. We measured the remapping detection thresholds for each combination of shift directions (left or right) and avatar appearances (realistic or abstract). The results show that realistic avatars increased the detection threshold (i.e., lowered sensitivity) by 31.3 percent than the abstract avatars when the leftward shift was applied (i.e., when the hand moved in the direction away from the body-midline). In addition, the proprioceptive drift (i.e., the displacement of self-localization toward an avatar) was larger with realistic avatars for leftward shifts, indicating that visual information was given greater preference during visuo-proprioceptive integration in realistic avatars. Our findings quantifiably show that realistic avatars can make remapping less noticeable for larger mismatches between virtual and physical movements and can potentially improve a wide variety of hand-remapping techniques without changing the mapping itself.