Requirement-Driven Design of Whole-Body Social Tactile Sensing via Virtual Human-Robot Interaction

Jul 16, 2026

Tactile sensing for social-physical human–robot interaction is often constrained by a hardware-first workflow, which couples sensor coverage, spatial resolution, and gesture recognition to fixed tactile configurations. The work introduces a requirement-driven framework that derives sensing needs—especially spatial resolution and sensor placement—from interaction data. A VR platform with haptic feedback collects high-resolution whole-body contact distributions across social scenarios, enabling identification of recurring touch gestures. Controlled studies produce an open dataset and quantitative baselines for skin coverage and sensor density on humanoid robots.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages8

Reading Time20 min

Introduction

Hardware-Driven vs Requirement-Driven Design

VR-Based Contact Data Collection

Dataset Construction and Gesture Identification

Deriving Sensing Requirements and Baselines

Requirement-Driven Design of Whole-Body Social Tactile Sensing via Virtual Human–Robot Interaction Dakarai Crowder1 , Ruohan Zhang 1 , Alexis E. Block2 , and Wenzhen Yuan 1 arXiv :2607 . 11690v1 [ cs .RO] 13 Jul 2026 Abstract—Tactile sensing for social-physical human–robot interaction (spHRI) is designed in a hardware-driven manner, where predefined sensor configurations constrain coverage, spatial resolution, and the range of recognizable gestures. We propose a requirement-driven framework that derives sensing requirements, specifically spatial resolution and placement, directly from interaction data. Using a VR-based platform with haptic feedback, we collected high-resolution whole-body contact distributions across multiple social scenarios, from which we identified nine recurring social touch gestures. Eight gestures were selected for controlled data collection with 18 participants, yielding an open-source dataset of 5,520 trials. Analysis of contact distributions and simulated tactile encodings provides quantitative baselines for skin coverage and sensor density on a humanoid robot platform. While demonstrated on a single robot platform, the methodology is designed to be transferable to other robot morphologies, potentially enabling morphology-specific sensing requirements to be derived prior to hardware fabrication. I. INTRODUCTION Robots are increasingly deployed in hospitals, restaurants, retail spaces, schools, and domestic settings, where they may interact directly with people [1], [2] . In these settings, natural and intuitive modes of communication can improve the efficiency, safety, and quality of human-robot interaction. Social touch is an underexplored modality in human–robot interaction. In human-human communication, touch plays a central role in conveying affect, intention, reassurance, social bonding, and functional cues [3] . Extending this capability to robots requires systems that can perceive and interpret touch-based interactions. However, achieving reliable social touch recognition remains a significant technical challenge. The design of tactile sensing systems typically follows a hardware-driven design paradigm: tactile sensor systems are first developed, and gesture recognition algorithms are subsequently built on top of the available sensing configuration [4], [5] . These systems often rely on large-area tactile skins to capture the spatial patterns of contact [6] . However, because whole-body tactile skins are expensive and require long customization cycles, hardware constraints limit sensing coverage and spatial resolution, and since gesture datasets are collected using these fixed configurations, recognition methods become inherently coupled to the sensors used during data collection. As a result, tactile sensor design for social touch is driven by engineering feasibility rather than the true requirements of natural interaction. 1 are with University of Illinois at Urbana-Champaign, Champaign, IL, USA {dcrowd3, rz21, [yuanwz](yuanwz}@illinois.edu)[}](yuanwz}@illinois.edu)[@illinois.edu](yuanwz}@illinois.edu) 2 is with Case Western Reserve University, Cleveland, OH, USA {[alexis.block](alexis.block}@case.edu)[}](alexis.block}@case.edu)[@case.edu](alexis.block}@case.edu) Fig. 1. Our framework collects collision data between a user and a virtual robot, enabling high-resolution contact extraction. (A) A user interacting with a virtual robot. (B) The user’s hand contact model decomposed into spheres and the convex hull of the robot’s upper arm mesh. (C) Sphere–triangle collision detection used to identify contact regions; circles indicate areas of contact. (D) Simulated sensor readings after processing the detected collisions. We propose an alternative, requirement-driven perspective: rather than starting with hardware, we begin by asking what tactile signals are required for natural social-physical human–robot interaction. We reconstruct human to robot contact independently of any

FAQs

What limitation of the traditional hardware-driven approach does the paper address?

It argues that starting from predefined sensor hardware limits coverage and spatial resolution, and tightly couples gesture recognition performance to the exact sensors used during data collection.

How does the proposed requirement-driven framework work?

It reconstructs human-to-robot contact independently of any tactile sensor configuration, then uses the reconstructed interaction data to derive sensing requirements such as spatial resolution and placement.

What data and datasets are created in the study?

The authors collect VR-based whole-body contact data across multiple social scenarios to form the spHRI Dataset, identify nine recurring gestures, and run a controlled collection for eight balanced gestures to produce the VR Gesture Dataset with 5,520 trials.

Jul 16, 2026

Upload by Valentina