Radar
Radar
A future collection domain.
Under developmentSuperpose research collection · 2026
Signal ATLAS
A map of the signal world
An evidence-led preview of a growing RF dataset collection for signal foundation models.
Explore the collection →Collection thesis
Signals need context, not just samples.
We pair RF I/Q with structured labels and provenance, then publish the validation boundaries alongside the capability. Verified natural-language annotations are the next modality—not a claim about what ships today.
Collection map
View all →Communications
Communications v1
A controlled communications signal capability spanning single emitters and wideband multi-emitter scenes.
Validated capability · Public corpus forthcomingCounter-UAS
Counter-UAS v1
Drone-emitted RF and co-band confusers for classification and scene understanding.
Validated capability · Public corpus forthcomingSatellite & Navigation
Satellite & Navigation
A future collection domain.
Under developmentSpectrum Environments
Spectrum Environments
A future collection domain.
Under developmentWireless Sensing
Wireless Sensing
A future collection domain.
Under developmentMultimodal path
Signal and structure
RF I/Q paired with machine-readable labels and generation provenance.
Verified language
Natural-language annotations grounded in structured scene truth and checked before release.
A broader signal map
Radar, satellite and navigation, spectrum environments, and wireless sensing—each only promoted when evidence supports it.
Validation principles
Evidence travels with the dataset.
- Separate generated evidence from real-world claims.
- Make limitations and non-claims prominent.
- Prefer established RF libraries and traceable provenance.
- Publish reports and representative figures at stable URLs.