Postdoctoral position on estimation of joint torques and contact forces from video in rock climbing

- Development of an experimental set-up for motion analysis in rock climbing from video
- Biomechanical modeling of the human body
- Development of an optimization-based Inverse Dynamics software

This postdoctoral position will explore new approaches in rock climbing motion analysis in order to estimate joint torques and contact forces from video. It follows previous works by Quaine et al. 2017 on an optimization method for simultaneous prediction of contact forces and joint torques from optical data. In Biomechanics, the usual experimental procedure to estimate muscle torques at joints requires to measure both kinematic data of the body from motion capture and contact forces with the environment from force sensors. While accurate, this approach is complex to deploy and limits the ability to address dynamics analysis in an environment such as rock-climbing mainly because of the absence of contact force sensors at each hold. Markerless methods are now expanding rapidly to allow an easier approach for kinematic analysis of body motion (OpenPose, Cao et al., 2017). In this project, we are interested in investigating further optimization procedure coupled with a markerless analysis in order to include estimation of joint torques and contact forces in the specific case of rock climbing.
Cao, Z., Simon, T., Wei, S.E., Sheikh, Y., Realtime multi-person 2d pose estimation using part affinity fields, CVPR 2017.

Quaine, F., L. Reveret, L., Courtemanche, S., Kry, P., "Postural regulation and motion simulation in rock climbing". The science of climbing and mountaineering, Routledge, pp.111-128, 2017, Routledge Research in Sport and exercise Science, 2017, www.motionworkshop.org/climbing_forces.mp4

The context of the project gathers the expertise of the Math-Sports group in Applied Mathematics and Computer Sciences at the LJK laboratory and the Biomechanics team at GIPSA laboratory, both in Grenoble area, France. The project will benefit from infrastructure offered at the laboratories, such as high-end motion capture system and 3D dynamic scanning. In addition, a close collaboration with the French National Team in Climbing (FFME) will allow to investigate the application of the new methods in the context of training of high-level athletes.

- Computer programming in C++ or Python
- Biomechanical modeling with software such as OpenSim or AnyBody
- 3D modeling and animation
- Computer Vision
- English