AVCR is a history-, geometry-, intervention-, and uncertainty-aware control and evidence platform for future cortical visual prostheses.
Preserve only the history and geometry needed for the current decision.
A useful percept is not determined by the current image alone. It may depend on the person, gaze, recent stimulation, cortical geometry, device state, task, and the reliability of the available evidence.
AVCR treats artificial vision as a closed-loop control problem: reconstruct the relevant state, certify the decision boundary, and preserve evidence before a future interface is allowed to act.
Phosphene location and quality may differ across people and sessions.
The same present command can have different consequences after different recent histories.
Gaze, maps, dropout, and interface conditions can alter spatial relationships.
Clinical translation requires explicit abstention, provenance, and replay - not only accuracy.
AVCR links scene information, history, geometry, device state, uncertainty, and feedback into a query-specific state before generating a bounded action proposal.
Extract the visual variables that matter for the current task.
Retain causal information that a snapshot representation would discard.
Respect person-, map-, and interface-specific spatial relations.
Gate every proposal by support, uncertainty, safety, and evidence requirements.
Use response and task outcomes to improve future authorized decisions.
The platform is organized around the decisions that must be defended: what information was used, what was uncertain, what action was authorized, and what evidence was preserved.
Encodes recent stimulation, response, and task history when it changes future decisions.
Causal stateMaintains spatial relations across gaze changes, map updates, and interface variation.
Structured mappingConverts novelty and ambiguity into limit, wait, or abstain behavior instead of silent overreach.
Safety policyCreates replayable logs, provenance, decision certificates, and hardware-in-the-loop evidence.
VerificationThe current implementation is deliberately non-human and non-stimulating. It is a physical hardware-in-the-loop bridge for testing software behavior, timing, faults, logging, and safety controls before any future sponsor-authorized interface.
AVCR is not presented as a clinical efficacy result. The current value is a formal architecture, an executable test environment, a bounded prototype path, and a prospective programme for stronger evidence.
The first commercial wedge is not a premature patient product. It is a non-clinical control, safety, HIL, and evidence package for teams developing or validating neuroprosthetic systems.
Need safer model-to-interface translation, reproducible experiments, and evidence-ready HIL.
Need a sponsor-specific control and evidence layer that complements hardware rather than replacing it.
Need pre-clinical decision logic, traceability, protocol-ready analytics, and explicit safety boundaries.
Need replay, provenance, fault evidence, cybersecurity scope, and decision-certification artifacts.
Anahita intends to use Abu Dhabi as AVCR's MENA operating base, with long-term founder presence, local technical capability, clinical-feasibility partnerships, and access to the life-sciences investment ecosystem.
Discuss an Abu Dhabi partnershipSet up the operating structure, founder relocation plan, IP and regulatory data room, and partner map.
Execute independently witnessed HIL work and recruit clinical, regulatory, engineering, and quality advisors.
Secure a scientific or clinical collaboration pathway and prepare the next regulatory and financing milestones.
Founder, inventor, and scientific lead
AVCR requires more than an image model. It sits at the intersection of control theory, bioengineering, medical-device software, uncertainty management, safety architecture, and clinical translation. The programme is founder-led and designed to expand through specialized engineering, clinical, quality, regulatory, and business-development partners.
Anahita welcomes conversations with life-sciences investors, hospitals, neurotechnology teams, implant-platform companies, research organizations, and independent verification partners.