Start from what happened.
The first trace is an observed REMUS H1 mission record. Its final SPEX score is 0.906 under the frozen evaluation protocol.
Research note · Equal-budget comparison
Can the same mission budget
measure more?
A real REMUS 100 mission trace and a frozen offline research candidate, evaluated over the same 28.394 km path budget. SPEX scores both routes; the planner proposes the candidate.
SPEX evaluates · The planner proposes
0.906Historical mission · SPEX score
0.974Research candidate · SPEX score
+0.068Absolute score improvement
28.394 kmEqual path budget
Observed source REMUS H1 navigation traceSPEX-derived scores and residual fieldsPlanner output offline geometric candidate
Reading the run
Mission complete is not objective proven.
Keep the budget equal.
The research candidate receives the same 28.394 km path budget. SPEX scores it at 0.974—an absolute improvement of 0.068.
Call the boundary.
The candidate is an offline geometric research result, not a route flown at sea or an operational vehicle-performance claim.
A result we did not hide
233.8 m looked better.
The run still did not win.
A mission-axis lawnmower baseline improved P95 residual from 236.0 m to 233.8 m, while mean and maximum residual worsened.
One metric improved Two deteriorated
Research access
Public evidence here. Guided dives by invitation.
This page remains a deliberately narrow public window—not a deployed capability or sea-trial claim. The interactive research environment is protected and used for scoped sessions with pilot, OEM and research partners.