My research asks how computational models can help an interface respond to the way a person perceives and acts. My current work focuses on tactile perception and physical feedback. I am developing this toward models that connect perception, action, and personalized design.
Tactile perception modeling
I develop computational approaches to understanding tactile perception, using psychophysical evidence to evaluate how well model predictions reflect human experience.
SwapSense
SwapSense: Reusable Vision-Based Contact Force Sensing Across Swappable Passive Haptic Modules
Accepted to UIST 2026
Minha Jeon, Yeseo Park, Inhyuk Song, Yilong Lin, Seungwoo Je, and Seungjae Oh.
Question. How can contact-force sensing be reused across interchangeable passive haptic modules, and how do those modules differ perceptually?
My contribution. I conducted multidimensional scaling (MDS) analysis to characterize the perceptual relationships among the haptic modules. This connected human similarity judgments with a spatial representation of their perceived differences.
Making Mid-air Button Contact Perceptible
Adaptive Feedback Model for Enhancing Interaction Saliency in VR
WHC 2025 Student Innovation Challenge
Inhyuk Song and Kihyun Park.
Question. How should a mid-air button adjust tactile feedback as a person adapts their pressing behavior?
Approach. I designed and implemented a prototype that adjusted the spatial trigger location of vibrotactile feedback using previous presses, then led an exploratory user study examining pressing behavior across feedback conditions.
My contribution. I led research design, implementation, evaluation, and analysis.
WHC 2025 Student Innovation Challenge
Peripheral signal representations
Comparing Peripheral Signal Representations for Predicting Perceived Intensity of Vibrotactile Stimuli
Presented at the Korean Haptics Conference, 2026
Inhyuk Song, Hee-Seung Moon, and Seungjae Oh.
This study compares simple spike-count representations with channel-weighted, compressed representations for predicting vibrotactile intensity. Using simulated tactile responses and published psychophysical data, it examines how the representation of peripheral signals affects predictions of human perception.
Spacewalk
Spacewalk Locomotion Technique and Tactile Feedback for Enhancing VR Experience while Lying Down
Korean Haptics Conference 2024 — accepted for poster presentation
Inhyuk Song and Seungjae Oh.
A VR locomotion prototype designed for interaction while lying down. The work explores a spacewalk metaphor, controller-based movement, and tactile feedback for interaction events. It provides a prototype and design proposal; evaluation of its effects on user experience is future work.
Korean Haptics Conference 2024 program
From perception to personalized interaction
Follow-up research in development
My next question is how physical feedback affects a particular person’s task success and input precision. I am developing an approach that connects perceptual modeling with biomechanical simulation and reinforcement learning, beginning with mid-air buttons.
The aim is to study how sensory information, movement constraints, and task goals shape behavior, and how a small number of observations might support personalization. Human experiments will be essential for evaluating both behavioral predictions and the designs those predictions suggest.
Undergraduate projects
Animal-Inspired Hand-Tracked VR Locomotion
Undergraduate team project · March–June 2024
A Unity/Quest prototype that mapped hand motion to bird-like locomotion. Tracked fingertip positions controlled the wings through two-bone inverse kinematics, while palm orientation and a finite-state controller governed movement. A One Euro filter smoothed the tracked fingertip positions.
RGB-D Scene Reconstruction and Proximity-Aware VR Guidance
Undergraduate team project · September–December 2023
A team prototype combining RGB-D scene reconstruction with visual guidance in VR. The project used an Azure Kinect and ORB-SLAM2 to reconstruct an indoor scene, processed point-cloud data into a mesh, and used distance-based coloring to highlight proximity to obstacles in Unity.
Mesh and Texture Post-processing for Text-to-3D Assets
Undergraduate team project · March–June 2023
Explored post-processing for meshes and textures generated with stable-dreamfusion. The project addressed coarse surfaces and texture noise through mesh subdivision, remeshing and smoothing, neural mesh denoising, texture upscaling, and color quantization.