Reconstruction of full-body joint kinematics from few, everyday sensors by combining sensor fusion with physics-informed learning.

Objectives:

1. Sparse-to-dense kinematics: full-body motion from a reduced set of wearable sensors.

2. Physics-informed, differentiable-biomechanics priors enforcing anatomically plausible motion.

3. Calibration-free and placement-agnostic estimation for everyday use.

4. Magnetometer-free operation via kinematic constraints (joint axes and centres).

5. Joint-angle accuracy assessed against optical ground truth (see Trustworthy estimation).