Postural stability in marine environments
OpenSim and SCONE workflow for movement reconstruction, joint-level analysis and controlled simulation, connected to Motion Induced Interruptions.
Geometric vision and 3D reconstruction
Camera calibration, normalised DLT, SVD, RANSAC-based fundamental matrix estimation and triangulation with reprojection analysis.
Radar, sonar and statistical detection
Academic track combining stochastic signal processing, Neyman–Pearson detection, covariance estimation, ROC analysis and Monte Carlo evaluation.
Quadcopter autonomy and estimation
IMU-based attitude estimation, Kalman filtering, PID control, Raspberry Pi–MultiWii integration and virtual waypoint missions.
Multivariable process control
Coupled level–temperature modelling from physical balances, followed by linearisation, state-space analysis and LQR synthesis.
Blind source separation with AMUSE
Latent-source recovery through lagged covariance matrices, generalised eigendecomposition and spectral validation under noise.
How projects are evaluated
The same critical framework is applied across domains.
What is being estimated?
The project begins with a measurable engineering question.
What supports the result?
Inputs, units, preprocessing and uncertainty are made explicit.
Why this model?
Method selection is justified against alternatives and assumptions.
What remains uncertain?
Failure cases and sensitivity are part of the final result.