SAFARI / APEX – Early wildfire detection via satellite and artificial intelligence

• Modular CubeSat platform : APEX enables the rapid integration of payloads and the reuse of the satellite for multiple missions.
• Onboard artificial intelligence (ODIN) : The ODIN system analyzes data directly in orbit and generates risk indicators based on spectral indices.
• Optimized transmission : Only relevant data is transmitted, reducing latency and communication volume.

SAFARI is a modular CubeSat platform (from 1U to 12U) integrating ODIN (Onboard Data Intelligence Engine), an artificial intelligence processing pipeline embedded onboard the satellite.

Where conventional approaches repatriate raw data to the ground for analysis, ODIN produces ignition risk maps directly in orbit from spectral indices (NDVI, NDRE, NIR, RedEdge), then transmits only the compressed results via a secure SDR link with post-quantum encryption. This inversion of the processing workflow drastically reduces the required bandwidth and accelerates risk detection.

The platform is designed to be reusable across multiple missions without redeveloping the onboard systems. For each subsequent operator, this translates into two concrete advantages: a structurally lower risk at initial commissioning and a reduced deployment cost.

• Latency under 20 minutes
• Fast onboard processing (~3 minutes)
• Data sovereignty
• Reduced mission costs 

• Wildfire management
• Environmental monitoring
• Defence and security
• Critical infrastructure

ProtoSat A-0.1 “Borealis” is operational. The CDR (Critical Design Review) was completed ahead of schedule, and the ODIN pipeline is validated in alpha on the Jetson Orin Nano.

The next major milestone is the CAN-SBX stratospheric balloon launch, planned for summer 2026 in partnership with SEDS Canada and the CSA. It will constitute the first validation of the system in near-space conditions.

Technology readiness level (TRL)

TRL 4-5: Laboratory validation is completed, and validation in a stratospheric environment is planned for summer 2026. The orbital launch of the prototype is targeted for 2028–2029, followed by constellation deployment.

The qualification of onboard AI algorithms for the space environment remains the central challenge: radiation tolerance, thermal cycling and latency constraints impose robustness standards far beyond the laboratory. In parallel, the selection and integration of the hyperspectral optical payload for the orbital platform are underway, with active discussions with Montréal-based partners. 

Access to environmental testing infrastructure adapted for CubeSats (vibrometry, thermal vacuum) also represents a concrete logistical challenge. Finally, defining the inter-satellite architecture and the DTC relay network will constitute a structural milestone for the transition to a constellation.

• The project primarily requires funding for the space qualification phase and ground segment development.
• Operationally, field partnerships with SOPFEU and NRCan would enable data validation under real -world conditions. Introductions to federal agencies—DND, NRCan, ISED— would pave the way for dual-use applications.
• Access to the CSA’s CUBICS program is targeted for the orbital launch of the prototype.
• Finally, industrial expertise in inter-satellite links (ISL) and space-grade computing would be decisive for further development.

• LASSENA Laboratory
• SEDS Canada
• Canadian Space Agency