Shared Materials - CSEDU 2026

Learning technical gestures without the expert guidance is challenging as books and videos are limited, especially in 3D information perception. While immersive room-scale Virtual Environments (VE) are often seen as an effective alternative, most are designed only for real-time practice based on imitating a 3D avatar demonstrating the gesture, lacking functionalities to help learners correctly position and orient themselves in the VE to observe and perceive all the required gesture features. To address this, the Automatic Observation Point Generator (AOPG) was designed to automatically generate the user’s head positions and orientations that enhance this perception. AOPG collects head motions from users freely navigating in a VE and applies a clustering-based process to generate relevant virtual camera positions and orientations. A first task involving lifting a heavy box, moving it, and placing it on the ground was considered. After collecting observation traces from 19 participants freely navigating in a VE, 11 observation points with high Average Silhouette Scores were generated from those traces. A survey completed by 36 new participants deemed 7 of these points highly appropriate for information perception. The study thus validates the potential of AOPG in this context, while keeping in mind that the system was designed to be adaptable to other types of gestures.