Evaluation of Retrieval Techniques for Out-of-Range VR Objects, Contrasting Controller-Based and Free-Hand Interaction

In VR environments like “sandbox” applications or interactive molecule simulators for education, physics-based objects can move beyond a user’s natural reach. We evaluate four interaction methods to support efficient and intuitive retrieval of objects for such events, including one that has not been evaluated previously and three others derived from well-known techniques. Considering the proliferation of camera-based hand tracking in VR headsets, there is a large interest in contrasting controller-based and free-hand interaction methods, so our evaluation considers both input types to further understand tradeoffs. We gathered performance data and subjective impressions from 52 subjects in a representative game-like puzzle task. A hand-extension (“go-go”-type) technique was least promising, with image-plane and pointing-based techniques being more promising. The recent tether-handle technique was roughly on-par with others for controller input (details vary by metric); it simply uses the same underlying grab metaphor as the main interaction. For free-hand interaction, its performance is reduced in a way that reflects broader problems of grab detection and manipulation methods for whole-hand interaction in VR.