Human plane
Intent + consent
Recommendations, evidence, uncertainty and consequences sit beside each person’s own response.
Shared intelligence, many physical realities
One commons. Every machine opens another adventure.
Carry models, identity, data formats and human controls between creations. Eyewear, food systems, machinery, tunnels, spacecraft and inventions still unnamed each bring distinct physics, relationships and consequences.
Common architecture
Compose the planes in any form.
Start with the person, a model, a controller, a sensor or an entirely different arrangement. The planes exchange evidence without collapsing into each other. Where physical hazards touch people, property or ecosystems, the risk picture shapes independent controls, diagnostics, redundancy, protection and safe states. Deterministic timing describes one property, rather than a complete safety case.
AI plane
Local inference
A chosen Arm or other processor runs models, language, vision, planning and the digital-twin interface.
Control plane
Risk-shaped control
Where physical hazards exist, real-time control and independent protection carry the limits, diagnostics and safe states revealed by hazard analysis.
Physical plane
Measured world
Sensors, actuators, power, radios, heat and failure are validated for the actual product environment.
XR wearables
Which wearable branch fits your experiment?
Raven Prism offers a direct-on-glasses prototype platform through its Linux foundation, 64-bit Arm compute, local apps and models, and third-party hooks.
The public raven-framework licence describes a source-available proprietary core rather than open source. Broader use, modification and distribution rest on a separate agreement. Raven’s starter project uses MIT; RavenOS, the core framework and hardware design sit under different terms.
| Path | What is promising | What is actually open | Aura role |
|---|---|---|---|
| Raven Prism | Linux underneath, Arm, local applications and models, privacy-oriented physical controls. | Starter app: MIT. Core framework: source-available proprietary. RavenOS and hardware design files are not publicly open. | Benchmark and integration reference. Dependence on the core begins with an explicit rights agreement. |
| MentraOS + Mentra hardware | Cross-compatible smart-glasses application stack; open hardware repository. | Public repositories identify MIT licensing for the OS and open-glasses work. | Strong open development path, with phone/runtime architecture constraints measured honestly. |
| Brilliant Labs | Developer-facing glasses and public hardware and software repositories. | Licence varies by repository. Reuse follows the terms attached to the exact Frame or Halo component. | Open-design comparator and possible field prototype platform. |
| Snapdragon AR platform | Purpose-built low-power vision/AI and distributed glasses-plus-host architectures. | Commercial platform and SDK; silicon design remains proprietary. | Performance and productisation reference for experiments beyond the present power and thermal envelope of open boards. |
Interfaces with bodies and shared space: open software and a processor describe parts of the wearable, rather than the safety of the whole. Optical and laser exposure, skin-contact heat, batteries, charging, radio-frequency exposure, Australian RCM compliance, accessibility and bystander camera or audio privacy each carry their own evidence and consequences.
In-home food computer
Compose a room-scale learning biosphere.
Begin with pantry, garden, fermentation, energy, water, inventory, harvest or another curiosity. Temperature, humidity, light, moisture, calibrated pH and electrical-conductivity instruments create different views of the living system. Local AI finds patterns and offers possibilities; each person follows, changes or rejects them. Food safety rests on validated processes and measurements rather than a sensor or model by itself.
Heating, pumps, dosing, pressure and food handling connect to validated process limits and the food-safety system relevant to the setting. Risk-shaped controls and physical fail-safes remain independent of the AI. Offline operation and chosen, consented sharing leave the data relationship with the household.
Food, health and laboratory interfaces
Different experiments meet different settings and responsibilities.
- Household food fermentation draws on established food processes, calibrated checks and hygiene evidence.
- FSANZ food-safety management rules attach to covered food businesses.
- Human biospecimens, pathogens and clinical experiments bring purpose-built laboratories, ethics relationships and biosafety oversight into the work.
- Australian medical-device rules attach according to the function and claims of software that diagnoses, treats, clinically monitors or recommends treatment.
- A microbiome sensor or model correlation begins a health question; it is not a health conclusion.
TGA: software-based medical devices ↗
NHMRC National Statement ↗
Industrial processing
Where does intelligence meet an independent safe state?
An Aura industrial node inspects material, schedules energy, detects anomalies and helps operators understand a process. Hazardous motion, heat, pressure, dose or release brings an independent protection relationship, so an AI fault does not become the hazard.
Industrial Arm computer
Runs Linux, vision, language, historian, digital twin and operator interface. Its failure leaves the safe state intact.
PLC / real-time controller
Carries deterministic sequences, alarms, physical limits and traceable control commands.
Independent safety system
Interlocks, emergency stops, guards, shutdown paths and a documented lifecycle arise from the machine risk assessment. A certified component describes itself, rather than certifying the whole machine.
Custom silicon at any moment: a sensor, security, control or wild-card accelerator offers another branch for simulation or fabrication. Machinery that touches workers or the environment brings a wider engineering and conformity question. The chip contributes evidence about itself, rather than certifying the machine around it.
Subterranean city
Open the underground as a field of questions.
Minjerribah is a sand island with regional and perched aquifers and groundwater-dependent ecosystems. A scenario layer opens exploration without physical disturbance. Any physical proposal enters living relationships with Traditional Owner governance, community purpose, geotechnical evidence, hydrology, ecology, planning and emergency systems.
A digital twin places non-intervention, distributed surface retrofit, shallow service corridors, speculative deep structures and new alternatives beside each other. No branch arrives as the default.
Questions that reshape geometry
- Whose Country, authority, knowledge and long-term benefit?
- Where do water, heat, smoke, waste and rescue routes go?
- Which forms remain inspectable, repairable and safely reversible?
- What surface alternative solves the same need with less disturbance?
- Which observations transform or end a concept?
Island network + space
Enter the network through any branch.
Architecture, orbital simulations, payload code, ground networks, commercial backhaul and fresh ideas unfold in parallel. Physical transmissions and launches meet the shared responsibilities of spectrum and orbit.
Local mesh
Solar-powered field nodes, local servers, open data formats and intermittent synchronisation connect sites whose owners and communities choose to host them.
Existing satellite backhaul
An available service links remote places while latency, outages, power and data priority become observations for continuing orbital design.
Ground / earth station partnership
Universities and recognised operators bring spectrum, antennas, scheduling, security and delay-tolerant networking into a shared experiment.
Hosted student payload
An isolated sensor and compute experiment rides on a partner mission. Radiation-tolerant command and recovery remain distinct from experimental AI.
Co-operative scientific constellation
Constellation design and simulation begin in the present. Physical deployment integrates launch, debris, spectrum, insurance, export, environmental and mission responsibilities through its operating network.