07 / ROADMAP
The concept becomes credible one controlled test at a time.
Build the pressure hull, track pods, bladder cells, sample bay, and buoy as separate test articles. Exercise seals, connectors, current limits, and retrieval points on the bench.
Run supervised tank and calm-water missions with no environmental claims. Compare stereo, sonar, and conditional green-LiDAR returns, and quantify false PI alerts.
Move to a permitted site with a written recovery plan. Compare retained cartridges against blanks and laboratory assays, then publish limitations alongside any exciting result.
Success is not a generated image of a nugget. It is repeatable localization, a known false-alert rate, safe recovery after simulated failures, a measured separation efficiency on seeded material, and a sample record another investigator can audit. The first release should publish its negative results and the conditions under which each sensor becomes unreliable.