ADVANCEMENTS IN THE ‘KINEPOSE’ FRAMEWORK
Full title: Advancements in the ‘KinePose’ framework for computer vision-aided reconstruction of pose and motion in computational human body models
Computational modelling using human body models (HBMs) can allow for parametric analyses of injury mechanisms and ‘what-if’ investigations into the potential of injury reduction strategies. Dynamics solvers for these models implement the equations of motion as an initial value problem. Understandably therefore, initially specified pose and motion of the model before impact, and active response during impact are important considerations, with significant effects on predicted injury patterns.
Acquiring motion data from volunteers in a laboratory environment is an option, but it presents significant challenges due to the ethical and practical constraints in exposing humans to dynamic stress scenarios. In contrast, while unstructured, there are various naturalistic data sources available, such as surveillance footage and vehicle-based sensors, that can provide real-world context. The applicant has already made significant progress in developing a pipeline involving computer vision, and kinematic optimisation techniques for inferring HBM pose and motion from video footage (KinePose).
The planned activities in the current project are twofold:
- Explore the potential of extending the KinePose pipeline for repositioning various HBMs (e.g. OpenSim, demoa allmin, VIVA+, the SAFER HBM, etc.) through a literature review on the state-of-the-art.
- Development of open-source tools for inclusion of pose and active response in these HBMs.
We expect to make contributions to open problems related to the representation of pose and motion in HBMs. Along with complimentary projects running in tandem (e.g., https://www.saferresearch.com/projects/surrogate-measures-safety-single-bicycle-crashes), we expect that this will kick-off further development and collaboration within the SAFER network. Additionally, all algorithms and tools developed through this project or subsequent initiatives will be released as open-source resources and made accessible to SAFER partners.
