Propulsion et navigation pour l'exploration sous-marine.

Underwater navigation is inseparable from propulsion. A control algorithm can only request motions that the vehicle, propellers and available thrust can physically produce in the surrounding water.

The project therefore connects propeller design and thrust testing with navigation programming instead of treating the mechanical and software work as independent tracks.

Propeller concepts need to be translated into testable behaviour. Thrust testing provides evidence for comparing designs and for understanding what the control system can reasonably command.

The useful result is not only a stronger component. It is a better model of the vehicle's available actuation, including the limits and asymmetries that navigation logic must respect.

Navigation programming depends on sensing, timing, coordinate assumptions, actuator response and the way the vehicle handles uncertainty. Each layer needs interfaces that make errors observable rather than silently passing them downstream.

This is where the project connects to my broader interest in physical intelligence: perception, decision and action only become useful when the whole loop can be evaluated in the real system.

Because the work is ongoing, this page records the engineering scope rather than claiming a final result. Future updates should add test methods, validated measurements and design changes only when they are ready to be made public.

That boundary is important. A credible project page should distinguish what has been attempted, what has been measured and what is still a direction for the next iteration.