Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session B: Wednesday, August 13, 1:00 – 4:00 pm, de Brug & E‑Hall

Inferring Virtual Height-Induced Changes in Postural Control via Inverse Optimal Control

Tahmineh A. Koosha¹, Fabian Hahne², Alap Kshirsagar², Nick Augustat, Christiane Melzig, Frank Bremmer, Jan Peters³, Dominik Endres¹; ¹Phillips-Universität Marburg, ²Technische Universität Darmstadt, ³German Research Center for AI

Presenter: Tahmineh A. Koosha

Humans adapt their postural control strategies in response to fear, but traditional sway metrics cannot directly reveal the underlying control objectives. We outline a preprocessing pipeline to enable inference of latent cost functions through our ongoing inverse optimal control (IOC) analysis. We exposed participants to ground (GC) and height (HC) conditions in virtual reality, while recording joint kinematics using Kinect-based motion capture. After aligning and denoising the data, we extracted joint angles (hip, knee, ankle) and computed summary metrics such as Mean, RMS, and Mean Power Frequency (MPF). Using Bayesian estimation, we found condition-dependent shifts in joint angle distributions, including reduced hip flexion and increased ankle stability under height. Our findings provide evidence of postural adaptation under perceived threat and lay the groundwork for the modeling of control strategies that govern balance in fear-inducing environments.

Topic Area: Methods & Computational Tools

Extended Abstract: Full Text PDF