The US manufacturing industry is currently facing a welding workforce shortage which is largely due to inadequacy of widespread welding training. To address this challenge, we present a Virtual Reality (VR)-based training system aimed at transforming state-of-the-art-welding simulations and in-person instruction into a widely accessible and engaging platform. We applied backward design principles to design a low-cost welding simulator in the form of modularized units through active consulting with welding training experts. Using a minimum viable prototype, we conducted a user study with 24 novices to test the system’s usability. Our findings show (1) greater effectiveness of the system in transferring skills to real-world environments as compared to accessible video-based alternatives and, (2) the visuo-haptic guidance during virtual welding enhances performance and provides a realistic learning experience to users. Using the solution, we expect inexperienced users to achieve competencies faster and be better prepared to enter actual work environments.

This work presents the design and evaluation of three bimanual interaction modalities for cross-workspace interaction in virtual reality (VR), in which the user can move items between a personal workspace and a shared workspace. We conducted an empirical study to understand three modalities and their suitability for cross-workspace interaction in VR.

In this work, we propose an end-to-end system that provides both hardware and software support for real-time gesture recognition. We apply a convolutional neural network over 3D rotation data of finger joints rather than over vision-based data, in order to extract high-level intentions (features) users are trying to convey. A pair of customized motion capturing gloves are designed with inertial measurement unit (IMU) sensors to obtain gestural datasets for network training and real-time recognition. A network reduction strategy has been developed to appropriately reduce a network's complexity in both depth and width dimensions while maintaining a high recognition accuracy with the classification model produced by the network. The classification model is able to classify new data samples by scanning a real-time stream of joint rotations during the use of the gloves. Our evaluation results expose the relationships between the network reduction hyperparameters and the change of recognition accuracy. Based on the evaluation, we are able to determine an appropriate version of the light network and achieve 98% accuracy.