From a trustworthy deterministic baseline to risk-aware inspection under uncertainty — where the work has been, and where it is going.
The deterministic baseline already flies the complete inspection trajectory — the fully-actuated base and the arm holding the camera on target. This is the ground the risk work perturbs.
Question → pre-registered prediction → measurement → conclusion. Two candidate mechanisms were eliminated by exact-zero tests before the surviving explanation was accepted.
Machine-checked against the community library, each proof shipping with a paper derivation: the determinant identity, the Lyapunov comparison theorem, LaSalle's invariance principle, and the passivity identity.
The mission-clock case is closed; the along-track velocity bias is the sharpened open question. Nothing uncertain is introduced until the deterministic baseline is locked.
The risk metrics that make degradation measurable. Three of four are implemented and verified.
Does risk-aware view scoring beat nominal and expected-value scoring under camera-pose uncertainty? The derivation is drafted and under review.
How should guidance behave when the state estimate itself is uncertain? Detailed after Phase A's gate.
Can the controller derate itself risk-awarely when its own actuation is uncertain? After Phase B's gate.
Reconstruction confidence and probabilistic margins as planning objectives. Research-first.
Planning and control for a free-flying space manipulator under state, actuation, and camera-pose uncertainty — measuring what caution costs and what it buys.
Two tracks run alongside and never gate the main line: the seven-degree-of-freedom extension and the Lean 4 formalization of the stability mathematics.