Digital mission card
Model the chosen environment, choose any scientific question and share the assumptions if the team wants others to reproduce or fork it. Run the same simulated dataset through competing algorithms.
Design far away. Learn close to home.
Two hundred observation focal points. Infinite ways to begin.
The supplied research sets a design challenge around about 200 focal points for Solar System observation. Students, universities, working engineers and AI copilots create, test and reshape the list while sensor packages, launch ideas and silicon architectures evolve in parallel.
The existing human network
This did not begin as a chip company.
It began as an invitation to think, model, build and learn together.
Luke describes the wider Maharashtra work as reaching several thousand students, politicians and parents. The supplied public SDNx post records the opening of an Innovation Lab in Kolhapur on 6 June 2019, names Luke Hayes Nathan among the invited technologists and announces an initial intake of 200 students across 12 aerospace, space robotics and automation projects.
That specific event record is shown below as user-supplied provenance. Crowd size and wider programme reach have not been independently audited for this site.
Open design challenge
Begin anywhere. Follow any question.
Use the draft catalogue as a launchpad, question it, combine entries, replace them or add your own. Work alone or with others. The prompts below offer a shared language for comparing and connecting projects. Use any that help, in any order, and add your own.
Focal pointWhere is the package going, and for how long?
QuestionWhat observable changes our understanding?
EnvironmentRadiation, heat, cold, dust, pressure, vibration, distance.
SensorsWhat does the team choose to sense, and with what range, resolution and calibration?
ComputeWhat happens locally, and what travels to Earth or elsewhere?
PowerEnergy source, peak load, sleep state and safe shutdown.
CommunicationsData volume, delay, antenna, optical/radio link and delay-tolerant networking (DTN) store-and-forward.
EvidenceCalibration plan, uncertainty, provenance and reproducible analysis.
FailureWhat happens when a sensor, bit, clock, link or model is wrong?
OpennessOpen interfaces, source, data licence and third-party restrictions.
Parallel learning branches
Flight ambition. Rapid iteration from simulation to silicon.
Choose any branch, run several in parallel, invent another, stop, fork or repeat. Every observation reshapes the direction as people acquire information and their needs evolve.
Bench sensor
Connect a real magnetometer, camera, spectrometer, radiation detector or environmental sensor to a small microcontroller (MCU). Calibrate it and log raw data in an open format when comparison matters.
Earth analogue
Test heat, cold, vacuum, vibration, radio loss or intermittent power with whatever safe equipment a team controls; add university or specialist facilities when their equipment is useful.
Hosted or university flight
Fly a self-contained experiment with a university CubeSat, hosted payload or other licensed programme. The team owns its inquiry while the host operates the shared command, safety and spectrum interfaces.
Resilient hybrid computer
A radiation-tolerant MCU handles command, reset and recovery while a higher-performance commercial off-the-shelf (COTS) Arm/FPGA device remains a replaceable experiment, supported by watchdogs, error-correcting code (ECC), current limiting and safe modes.
Small-batch silicon experiments
Student teams model any architecture immediately and send small designs through open-PDK, Tiny Tapeout or MPW routes whenever useful. Accessible fabrication often uses mature nodes. Leading-edge silicon and silicon tested for flight environments explore different physics, costs and mission questions in parallel.
Space-hardened does not mean smallest
Choose the node for the mission.
Space electronics face total ionising dose, displacement damage and single-event effects. Smaller geometries improve performance and sometimes dose tolerance, while dense, low-voltage logic sometimes becomes more sensitive to single-event upsets.
Radiation-hard-by-design draws on redundancy, error correction, spacing, isolation, watchdogs and recoverable safe modes. Radiation evidence belongs to the actual part and mission environment, rather than an Arm logo or node label.
ESA: radiation-hardness at scaled nodes ↗
NASA/JPL: radiation effects ↗
One architecture to remix
Supervisor + experimental payload pattern
Question, fork, replace or invert any part as mission evidence evolves.
- A radiation-tolerant MCU handles boot, command, reset and safe mode.
- A commercial Arm SoC or FPGA runs experimental AI workloads.
- Independent power domains isolate and restart after a fault.
- Checksums, ECC, scrubbing and duplicated observations detect or correct defined fault classes, but not every possible corruption.
- Mission- and part-specific testing covers total ionising dose, displacement damage and single-event effects.
- AI runs freely in simulations and isolated payload experiments; operational flight recovery remains independently controllable.
NASA: resilient affordable CubeSat processor concept ↗
Microchip SAMRH71 radiation-hardened Arm MCU ↗
A shared observatory
A space-weather learning hub.
Alongside Solar System swarm design, teams fuse existing public feeds with small local sensors and digital twins.
The supplied notes already sketch magnetometers, energetic-particle detectors, solar-wind plasma measurements, imaging and spectroscopy; event rules; data provenance; XR views; and delay-tolerant storage. That is a strong curriculum when each claim is classified and every alert has an accountable source.
Observe
Ingest solar images, solar-wind conditions, magnetic fields, particle fluxes, ionospheric measurements and local instrument health.
Compare
Run baseline forecasts beside student models. Preserve inputs, software version, uncertainty, output and later observation.
Challenge
Let alternative hypotheses compete on the same data. Pair beautiful visualisations with predictive scoring, uncertainty and an honest record of misses.
Open earthquake challenge
Explore every signal. Publish the evidence trail.
The US Geological Survey reports that no method has yet demonstrated reliable prediction of a major earthquake’s time, location and magnitude. That published baseline is part of the challenge, not the edge of it. Teams explore space-weather, radon, animal-behaviour, ionospheric, geological and yet-unnamed correlations; publish hypotheses, methods, forecasts, misses and falsification records; and let competing models meet the same observations.
Operational early warning is a different signal. It detects rupture after it begins and alerts places before strong shaking arrives. Anyone presenting a service as an official public emergency warning states its operator, evidence, uncertainty, track record and accountability. Experimental signals remain free to be shared as experimental signals.
What the AI is for
A thousand copilots widen imagination. Shared tests turn imagination into knowledge.
People gathering around a package
Work alone, with AI or with any mix of peers and specialists. Choose companions from this list, elsewhere or nowhere at all.
- Student designer owns the question and explains every trade-off.
- AI copilots generate alternatives, simulations, code and critique.
- University reviewer challenges the science, method and laboratory conditions.
- Systems mentor stress-tests power, interfaces, failure and manufacturability.
- Data steward traces provenance, uncertainty, consent and licence.
- Independent red team tries to falsify the claim and break the design.
- Cheerleader, guide or advocate grows confidence, joy, access and momentum.