InLocoMotion: Dynamic 3D human body shapes from static 3D scans and sparse motion tracking for the improvement of human-product systems: a case on cycling drag force estimation

The human body is a complex biomechanical system with a large anatomical diversity. New methods for industrial design are emerging based on accurate 3D models and statistical analysis of their rich spatial geometry and complex variations. Most applications of this 3D anthropometry in the field of Product Development are confined to static 3D shapes, whereas many products such as garments, (space) suits, sports equipment, medical devices, vehicles, and household appliances might benefit from accurate dynamic deforming 3D models of the human body. Currently, even for products that dynamically interact with the human body (e.g. shoes), only static geometric information is considered, thereby ignoring the potential to consider full 3D surface in motion and dynamic deformation. In this Baekeland PhD project, we will construct and validate design methods to use dynamic 3D anthropometry in the process of product development and extend the use of static 3D anthropometry. We will combine the aforementioned state-of-the-art statistical shape models with state-of-the-art animation techniques and translate them to CAD tools and techniques to support the envisioned extension. Firstly, a method is provided to generate any individual 3D body shape in any position from a combination of geometrical shape information and temporal position parameters that is both easy to assess. Shape information will comprise an individual's shape in a static pose, e.g. standing position, or a set of 1D anthropometric parameters. Position parameters will be achieved by adapting reliable and accurate of-site motion capturing techniques. We will also investigate how product developers might use these parameterized person-specific dynamic 3D models in the process of product development i.e., what shape and position information they need during the design process and what the requirements are on that information such that they will use it most effectively. This will pinpoint how product developers will preferably interact with the envisioned human-product models. Next, these requirements will be used to develop CAD tools and techniques in which products can be designed on person-specific dynamic human body models, and resulting human-product models can be tuned and optimized by a anthropometric measurements and position parameters. For instance, a stack of person-specific human-product models can be generated with the same effort required to generate only one such model. Finally, we will validate our method by simulating drag force of cyclists, in comparison to ground truth values in a wind tunnel. The target is to come very close to real drag force values with a fraction of the cost and investment. Although this PhD will directly contribute to the subfield of aero-design and engineering in cycling, the lead up methods will also prove the accuracy of underlying models. We will thus establish a direct and accurate link for the product developer between human(-product) CAD models and the actual physical model to support simulation, verification and validation. Our method will improve the process of product development in several aspects. It will have the potential to reduce development costs by omitting the need for physical prototyping. An early stage verification of product functionality and composition of design specifications will require less iterations and entail a shorter time to market for new products. Our method will not only enhance comfort and functionality of final products but will also allow to develop new categories of consumer and medical wearable products, that owe their functionality to close and dynamic product-body interaction and extensive ergonomics.

Keywords:3D SCANNING, 3D ANTHROPOMETRY, LOCOMOTION

Disciplines:Biomechanics

Project type:Collaboration project