Bend It, Aim It, Tap It: Designing an On-Body Disambiguation Mechanism for Curve Selection in Mixed Reality CCS: Human-centered computing Virtual reality CCS: Human-centered computing Mixed / augmented reality CCS: Human-centered computing Empirical studies in HCI CCS: Human-centered computing Inter

Xiang Li, Per Ola Kristensson · 2025

Read Bend It, Aim It, Tap It: Designing an On-Body Disambiguation Mechanism for Curve Selection in Mixed Reality CCS: Human-centered computing Virtual reality CCS: Human-centered computing Mixed / augmented reality CCS: Human-centered computing Empirical studies in HCI CCS: Human-centered computing Inter by Xiang Li, Per Ola Kristensson (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-13748.txt.

Introduction

Li and Kristensson design new ways to pick objects in mixed reality in twenty twenty-five. Picking an object is hard when a scene is dense or items hide behind others. A straight ray from the hand often points at several items at once. The authors offer two linked techniques to fix this.

One technique bends the ray into a curve guided by how much the finger bends. The other projects the four closest candidate objects onto the forearm of the user. Twenty-four people took part in a study that crossed both factors. The on-body display cut errors and physical strain, and users liked it more. The curved input reached hidden targets well but could take longer and more effort.

Catalog record

CREATORXiang Li, Per Ola Kristensson
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE53 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-13748

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Bend It, Aim It, Tap It: Designing an On-Body Disambiguation Mechanism for Curve Selection in Mixed Reality CCS: Human-centered computing Virtual reality CCS: Human-centered computing Mixed / augmented reality CCS: Human-centered computing Empirical studies in HCI CCS: Human-centered computing Interactive systems and tools Xiang Li, Per Ola Kristensson Abstract . Object selection in Mixed Reality (MR) becomes particularly challenging in dense or occluded environments, where traditional mid-air ray-casting often leads to ambiguity and reduced precision. We present two complementary techniques: (1) a real-time Bézier Curve selection paradigm guided by finger curvature, enabling expressive one-handed trajectories , and (2) an on-body disambiguation mechanism that projects the four nearest candidates onto the user’s forearm via proximity-based mapping. Together, these techniques combine flexible, user-controlled selection with tactile, proprioceptive disambiguation. We evaluated their independent and joint effects in a 2 × 2 2\times 2 within-subjects study ( N = 24 N=24 ), crossing interaction paradigm (Bézier Curve vs. Linear Ray) with interaction medium (Mid-air vs. On-body). Results show that on-body disambiguation significantly reduced selection errors and physical demand while improving perceived performance, hedonic quality , and user preference. Bézier input provided effective access to occluded targets but incurred longer task times and greater effort under some conditions . We conclude with design implications for integrating curved input and on-body previews to support precise, adaptive selection in immersive environments. Figure 1. Overview of our 2\times 2 study design, crossing two interaction media (Mid-air and On-body) with two interaction paradigms (Linear Ray and Bézier Curve) ( Li and Kristensson 2025 ) . We investigate how the interaction medium and interaction paradigm jointly affect selection performance, effort, and user experience in virtual environments. A 2×2 study design with two interaction media—Mid-Air and On-Body—and two interaction paradigms—Linear Ray and Bézier Curve. Illustrations show a VR user performing selection either in mid-air or on their forearm, using either a straight ray or a curved Bézier trajectory. 1. Introduction Object selection is a fundamental interaction task in Mixed Reality (MR), yet it becomes particularly challenging in visually dense or spatially occluded environments. Mid-air pointing techniques, such as Linear Ray-casting ( Bowman and Hodges 1997 ; Pierce et al. 1999 ) , are widely adopted due to their intuitiveness and compatibility with distant targets. However, they often suffer from selection ambiguity, limited precision, and a lack of tactile feedback, specially when multiple targets are closely spaced or partially occluded ( Jr et al. 2017 ) . As MR applications scale to more complex scenes, there is a growing need for selection techniques that offer both flexibility and disambiguation support. To address these limitations, prior work has explored two promising directions. The first involves enhancing the expressiveness of the selection path. Curved trajectories, such as Bézier curves, offer users the ability to reach around occluding objects or select targets in cluttered spaces without physically repositioning ( Olwal and Feiner 2003 ; Steinicke et al. 2004 ; Lu et al. 2020 ) . For example, the Flexible Pointer ( Olwal and Feiner 2003 ) enabled Bézier Curve-based selection by combining two-hand input to modulate curve shape, though it required complex hardware coordination. The Improved Virtual Pointer ( Steinicke et al. 2004 ) took a system-driven approach, bending rays toward nearby targets using proximity, but reduced user control over the trajectory. While these techniques demonstrated the value of curved input, they either lacked user-driven control or imposed cumbersome setups. Our preliminary work ( Li and Kristensson 2025 ) also explored Bézier-based selection with on-body projection but only proposed the initial design with . This paper substantially extends that work with a full implementation and a comprehensive factorial user study. The second direction leverages the user’s body as an interaction surface ( Harrison et al. 2010 ; Gustafson et al. 2011 ; Yu et al. 2022 ) . On-body input offers passive haptic feedback and proprioceptive grounding, which can improve selection clarity and reduce cognitive load ( Simeone et al. 2014 ) . Yet, on-body disambiguation has rarely been integrated with curved input, and it remains unclear whether a single on-body strategy can support both expressive and conventional pointing paradigms. In this paper, we bridge two strands of MR selection research (i.e., trajectory expressiveness and target disambiguation) by introducing an integrated system that combines real-time curved input with tactile, proprioceptive support. We propose two complementary techniques: (1) a Bézier Curve selection paradigm driven by index finger curvature, enabling expressive, user-controlled trajectories with one hand; and (2) a proximity-based on-body projection mechanism that maps the nearest selectable targets onto the user’s forearm, allowing direct disambiguation via the non-dominant hand. The number of projected candidates (i.e., 4) is motivated by physical constraints of the forearm and prior findings on on-body menus ( Li et al. 2024a ; Azai et al. 2018 ; Tsimbalistaia et al. 2025 ) To evaluate their effectiveness and interplay, we conducted a controlled user study ( N=24 ) using a 2\times 2 within-subjects design. We varied Interaction Paradigm (Linear Ray vs. Bézier Curve) and Interaction Medium (Mid-air vs. On-body), resulting in four interaction conditions. This factorial design allowed us to examine the independent and combined effects of curve-based input and on-body disambiguation. Our results show that on-body disambiguation significantly reduces selection errors, lowers physical demand, and improves perceived performance and user preference across paradigms. The Bézier paradigm enables flexible access to occluded targets but introduces trade-offs in task time and physical effort, especially in on-body conditions. Together, these findings reveal the synergies and tensions between on-body disambiguation and trajectory expressiveness in MR selection. Our contributions are as follows: • We introduce a real-time Bézier Curve generation technique driven by finger curvature, enabling expressive and controllable curved selection in immersive environments. • We propose a proximity-based on-body projection mechanism that supports tactile disambiguation by mapping up to four nearby targets onto the user’s forearm. • We empirically demonstrate that on-body disambiguation improves accuracy, reduces workload, and enhances user preference, while Bézier input offers spatial flexibility with measurable trade-offs. 2. Related Work 2.1. Body-Centric Interaction On-body interaction techniques use the human body as an interactive platform, significantly enhancing immersion and proprioception in virtual reality (VR) environments ( Harrison et al. 2012 ; Bergström and Hornbæk 2019 ; Coyle et al. 2012 ; Mueller et al. 2021 ; Mueller et al. 2023 ) . These approaches improve accuracy in eyes-free targeting, owing to the natural familiarity of users with their own body dimensions ( Weigel et al. 2017 ; Gustafson et al. 2010 ) . Extensive research has been conducted on utilizing various body parts (e.g., arms, palms, skin) as interfaces for VR interactions ( Chatain et al. 2020 ; Dezfuli et al. 2014 ; Mistry et al. 2009 ; Wang et al. 2025 ) . For example, Skinput ( Harrison et al. 2010 ) and Touché ( Sato et al. 2012 ) demonstrate advanced gesture recognition capabilities by detecting acoustic and capacitive signals directly from the skin. Further studies have assessed the effectiveness of interfaces anchored to the non-dominant arm for precise pointing tasks in VR, as explored by Li et al. ( Li et al. 2021 ) . The development of body-centric selection and manipulation techniques such as the Hand Range Interface ( Xu

Read the complete text