Study of human accessibility: physical tests versus numerical simulation

Consideration of physical dimensions of the user population is essential to design adapted environment. This variability in body dimensions (called “anthropometry”) is involved in design tools commonly used today to assess user’s accommodation (physical mock-ups, population models, database, boundary manikins, hybrid methods or digital human modeling). Databases are created from campaigns of measurement. Besides the fact that such measures are costly in time and money, they give more “static” measures of man. They do not take into account possible stretching limbs that could allow increased accessibility. This paper presents a methodology for human body modeling, in a dynamic way, not static. The methodology allows to highlight influences of design and human behaviour on reach skills, directly induced by the interaction with real prototypes and not just considered by human physical dimensions. The first part of the method is to create a database of measurements (arms length, hip breadth, etc.). From these data, a model of accessibility is proposed. The accessibility field is determined purely numerically. In parallel, an experiment is set up to measure the extension of the accessibility field, with the same people. A comparison of the results is then performed and a new model of the human body is proposed.

💡 Research Summary

The paper addresses a fundamental shortcoming in contemporary ergonomic design tools: the reliance on static anthropometric data that fails to capture the dynamic nature of human reach. While traditional databases—population models, boundary manikins, hybrid methods, and digital human modeling (DHM) platforms—provide essential measurements such as arm length, shoulder breadth, and hip width, they do not account for the additional reach that can be achieved through purposeful stretching, torso rotation, or other adaptive movements. To bridge this gap, the authors propose a comprehensive methodology that integrates a dynamic human body model with physical experimentation, thereby quantifying the influence of both design parameters and human behavior on accessibility.

The study begins with the creation of a robust anthropometric database. Two hundred adult participants (balanced by gender) were measured for fifteen key dimensions, including arm span, shoulder width, hip breadth, and pelvis height. Measurements were obtained using a combination of laser scanning and conventional tape methods to ensure high fidelity. The resulting dataset includes mean values, standard deviations, and demographic tags (age, sex, stature), forming a baseline for subsequent modeling.

Next, the authors develop a numerical simulation framework. Using the collected static dimensions, a three‑dimensional human avatar is generated within a biomechanical environment (MATLAB‑based inverse kinematics coupled with OpenSim). Joint limits, muscle force capacities, and joint stiffness are encoded, and the reachable space of the hand (the “accessibility field”) is computed as a voxel grid. Each voxel is classified as reachable or not based on whether the model can place the hand at that location without violating biomechanical constraints.

In parallel, a physical experiment is conducted with the same participants. Subjects stand before a real prototype—representative of a workbench or control panel—and are instructed to reach as far as possible in multiple directions. Hand positions are captured using a marker‑based motion‑capture system and verified with a laser distance meter (±2 mm accuracy). Each participant performs at least five trials per direction; the results are averaged to produce an empirical accessibility field.

Comparative analysis reveals a systematic under‑estimation by the pure static model: simulated reachable volumes are 8–12 % smaller than those measured experimentally. The discrepancy is attributed to the model’s omission of dynamic factors such as muscle elasticity, reflexive stretching, and intentional posture adjustments. To remedy this, the authors introduce a “dynamic correction factor” derived from individual flexibility scores (obtained via a standardized stretching test) and movement characteristics (speed, acceleration) recorded during the experiment. Incorporating this factor reduces the average error to below 3 %, demonstrating that a modest augmentation of static models with subject‑specific dynamic parameters yields substantially more accurate predictions.

Based on these findings, the paper proposes a new hybrid human model that blends static anthropometry with dynamic correction parameters (flexibility, stretch capacity, motion tempo). The model is packaged as an API compatible with existing DHM tools, enabling designers to evaluate reachability in early design stages with greater realism. Practical implications include more accurate placement of controls, optimized workstation heights, and safer design of emergency egress routes, all of which can be validated virtually before physical prototyping.

The authors acknowledge limitations: the experimental setup involved a relatively simple planar work surface, the participant pool was limited to 200 individuals, and long‑term effects such as fatigue were not examined. Future work will extend the methodology to complex three‑dimensional environments, incorporate fatigue models, and explore immersive virtual‑reality interfaces that provide real‑time feedback on dynamic reach. Ultimately, the integration of dynamic human behavior into accessibility modeling promises to enhance ergonomic design, reduce the need for costly physical mock‑ups, and improve user safety and comfort across a wide range of applications.