Experimental validation of a computer-vision based method to assess the aerodynamic drag of cyclists

Abstract

Reducing drag is a major challenge in cycling. In fact, it is a well-known fact that, on flat road conditions, aerodynamic drag represents about 80% of the total resistive forces applied to the cyclist. The aerodynamic drag is given by the following equation: , where  is the air density,  the cross-sectional area or frontal area and,  , the drag coefficient. In order to reduce the aerodynamic resistive forces, one search to minimize , which requires to be quantified. In the literature, different methods have been proposed; we can mainly quote: wind tunnel [1], dynamometric measurement [2], deceleration [3], linear regression [4], and 3D digitalization-based method [5-7].

We previously introduced a new computer vision-based method to assess the aerodynamic drag of cyclists [8,9]: first a 3D+t model of the cyclist and his bike is built and thereafter this model is processed by a CFD solver to assess the aerodynamic resistive forces. This method offers a low cost alternative to the wind tunnel measurements and does not involve any special infrastructure (track) like the linear regression technique. Moreover, it overcomes the limitations of the classic «3D + CFD» methodologies that we investigated in a precedent work [8]. In previous work, we also evaluated performance of our method [9-10]. In this work we create a new dataset recorded in a wind tunnel and use it to experimentally validate our method.