SIM2VR: Towards Automated Biomechanical Testing in VR

自动化生物力学测试在开发虚拟现实应用方面具有巨大潜力，因为可以在设计过程的早期阶段通过计算机模拟获得有关用户行为的初步洞察。特别是，它可以预测用户的运动和人体工学变量，例如疲劳，在进行用户研究之前。然而，存在一个根本性的断层，即承载最先进的生物力学用户模型的模拟器与用于开发和运行虚拟现实应用的模拟器之间存在巨大差异。现有的用户模拟器经常难以捕捉真实世界虚拟现实应用的复杂性和细微差别，从而降低了用户预测的生态效度。在本文中，我们介绍了SIM2VR，这是一个系统，通过在两个过程之间建立连续闭环，将用户模拟与给定的虚拟现实应用程序对齐。这是首次在真实用户交互的相同虚拟现实应用程序中直接训练模拟用户。我们证明SIM2VR可以预测快节奏、动态街机游戏中用户表现、人体工学和策略的差异。为了将自动化生物力学测试的范围扩展到简单的视觉运动任务之外，需要在认知模型和奖励函数设计方面取得进展。

SIM2VR: Bridging the Gap between Biomechanical Simulation and Virtual Reality Applications

SIM2VR establishes a continuous closed loop between user simulation and a given VR application, enabling training simulated users directly in the same VR application that real users interact with, thus improving ecological validity of user predictions.

SIM2VR can predict differences in user performance, ergonomics and strategies in a fast-paced, dynamic arcade game. The system has potential for the development of VR applications by gaining initial insights into user behavior early in the design process. Advancements in cognitive models and reward function design are needed to expand the scope of automated biomechanical testing beyond simple visuomotor tasks.

Related work includes studies on biomechanical simulation, virtual reality applications, and user behavior prediction in VR environments. Some relevant papers include 'A Biomechanical Model for the Analysis of Human Movement and Its Application to the Design of a Gait Rehabilitation Device' and 'Predicting Human Behavior in Virtual Reality'.