Menu

AI modules and carrier boards

Two DM3 modules share one form factor and one set of carriers. DMCM5 fits platforms already built on RPi CM5.

Which module fits your platform?

DM3-32DM3-8DMCM5
AI performance32 TOPS8 TOPS6 TOPS
CPU8× Cortex-A722× Cortex-A728-core Arm
Real-time cores4× Cortex-R5F6× Cortex-R5F—
RAM16 / 32 GB4 / 8 GBup to 8 GB
Storage32 GB (up to 64)32 GB (up to 64)up to 64 GB
Temperature−40 … +125 °C (junction)−40 … +85 °C (junction)—
Form factor80 × 55 mm80 × 55 mm55 × 40 mm, RPi CM5
CarrierCarrier-1, Carrier-2Carrier-1, Carrier-2Any RPi CM5 carrier
Best forHeavy autonomous systems, multi-cameraUGV controllers, light AI payloadsPlatforms already built on RPi CM5

Which carrier takes which module?

DM3-32 and DM3-8 share one footprint, so both fit every DM3 carrier. Swap the module without redesigning the carrier.

Carrier-1Carrier-2 AvionixRPi CM5-compatible carriers
DM3-32YesYes—
DM3-8YesYes—
DMCM5——Yes

Prices for partners

Sign in to configure a module and see indicative prices for your quantity.
Have a Google account? Sign in with one tap. No Google account? Use any other email below.
We will email you a sign-in link. Open it from your inbox and the price calculator will appear.

Frequently asked questions

Which flight controllers work with DroneMate AI modules?
Carrier-2 Avionix has the signals for Cube Orange+ / Blue / Red and Pixhawk v5x / v6x already routed on the board: UART, CAN, Ethernet and GPIO. DM3-32, DM3-8, Carrier-1 and DMCM5 connect to other flight controllers by cable over UART, CAN or Ethernet.
Which autopilot firmware is supported?
ArduPilot and PX4, both talking to the AI module over MAVLink. ArduPilot is the planned default, and PX4 can be installed instead.
What is the difference between DM3-32 and DM3-8?
DM3-32 delivers 32 TOPS with 8× Cortex-A72, 4× Cortex-R5F and 16 / 32 GB RAM. DM3-8 delivers 8 TOPS with 2× Cortex-A72, 6× Cortex-R5F and 4 / 8 GB RAM. Both use the 80 × 55 mm footprint and the same carriers, so you can start on DM3-8 and move to DM3-32 without changing the carrier.
Which carrier takes which module?
DM3-32 and DM3-8 fit both Carrier-1 and Carrier-2 Avionix. DMCM5 fits RPi CM5-compatible carriers.
Which AI frameworks can I run?
DM3-32, DM3-8 and Carrier-1 come with a Linux image, a setup guide and a guide to converting neural networks for the NPU. Supported runtimes are ONNX Runtime, TensorFlow Lite Runtime and TVM / DLR.
How do I get datasheets and integration documents?
Public specifications are on each product page. Full datasheets and integration documents are available under NDA to engineering teams integrating our modules.
How do I start?
Book a 30-minute technical call. Next comes bench integration; once the bench setup works, we configure the module and carrier for your vehicle as a pilot batch.
Does it work without an internet connection?
Yes. AI inference runs locally on the board, without an internet connection or a cloud service.

Product Details - Request

To receive specifications or technical consultation, please leave your contact details below.