Introductory article
When a model fails without hardware failure
Why validity of system knowledge must be treated as a resource.
Failure without broken electronics
Imagine a robot whose sensors and motors still work, but its mathematical model no longer predicts its behavior. Healthy hardware cannot guarantee a safe decision when the model or estimator is wrong.
The dissertation therefore introduces an epistemic resource: a model, estimator or algorithmic representation whose validity contributes to system capability.
Two resource classes
\[
\mathcal R=\mathcal R_{phys}\cup\mathcal R_{epist}
\]
- Rphys: sensors, actuators, processors and power sources.
- Repist: mathematical models, estimators and learned algorithms.
The framework checks not only whether a component is available, but also whether its data and predictions remain trustworthy. If validity falls, the supervisor must reconsider which configuration is admissible.
Continue with the transformation space and trajectory planning.