Clear status. Honest limits. Useful outcomes.
Explore the problem each project addresses, its current stage, and the evidence needed to move it forward. Public descriptions explain purpose and evaluation goals; proprietary implementations and sensitive material are not included.
ScubaRC Genesis · Community Compute & Commonwealth Surf
Question: Will people voluntarily participate, and can a browser return a result that is checked and saved?
Available now: A free browser game with server-verified scores and a separate, opt-in synthetic compute task with result receipts. These demonstrate participation and infrastructure; they do not establish scientific utility or a count of unique human participants.
Engineering discipline: Genesis uses immutable release candidates, independent admission, server-authoritative verification, adversarial timing/security testing, production rollback controls, real-device QA, and a formal post-mission lessons-learned process.
Next evidence needed: Completion and failure rates across devices, repeat participation, and a suitable public-data workload chosen with a domain researcher.
Responsible AI, Robotics & Human-Machine Systems
Research and development in advanced robotics, physical AI, human–robot teaming, secure AI adoption, model evaluation, and accountable automation. We investigate how perception, field computing, and human oversight can work together in physical environments.
Research question: Can AI-supported work preserve context and accountability across interruptions, devices, and changing conditions? Evaluation should examine task completion, errors, recovery, and the human effort required—not just fluent output.
Perception, computation, and human judgment in the physical world.
ATHENA and ARES connect ScubaRC’s work on sensing and field computing to advanced robotics. Application areas under investigation include remote inspection, environmental observation, operator assistance, and human-supervised work in difficult environments.
Key research questions include how to communicate uncertainty, maintain useful operation when connectivity degrades, and give people clear control over machine actions. Evaluation needs to measure perception errors, response time, recovery, and operator workload.
These are development directions, not claims of a deployed autonomous robot or a validated safety-critical system. We welcome robotics, controls, sensing, and human-factors collaborators to define bounded demonstrations and meaningful benchmarks.
Explore ATHENA, ARES, and related researchCognitive Security & Information Integrity
Defensive study of influence operations, narrative manipulation, synthetic personas, credibility laundering, AI-amplified persuasion, and decision-layer resilience.
Research question: Can analysts recognize manipulation while retaining source context, uncertainty, and competing explanations? Evaluation should use documented cases and measure false alarms as well as useful findings.
Quantum & Advanced Sensing
Public-safe research into inspection, magnetometry, signal extraction, anomaly visualization, and forensic traceability.
Research question: Can complementary sensing methods help identify areas that warrant closer inspection? Proposed evaluations require reference measurements, known defects, and uncertainty reporting. Inspection concepts do not replace established inspection methods or imply a validated field instrument.
OSINT, Early Warning & Public Safety
Responsible methods for identifying weak signals, anomalies, coordinated activity, and emerging risk in public data while protecting sources, users, and civil liberties.
Research question: Can public information reveal meaningful early warning without overwhelming people with noise? Evaluation should examine lead time, source quality, false alarms, and privacy impacts.
Make the evidence useful to a reviewer.
We distinguish proposed work from completed work, prototypes from products, and hypotheses from validated findings. Public materials describe problems, principles, methods, and results at a level appropriate for responsible review.
Bring a question we can test together.
Useful starting points include a public dataset, a published baseline, an independently checkable calculation, or a classroom activity about verification and uncertainty.
We welcome help defining success measures, reviewing limitations, and selecting a small scientific workload. Collaboration begins with a bounded question and agreed disclosure terms; institutional partnerships are not implied.
Start a research conversation