Nvidia Orin NX
Information about the onboard Jetson
Last updated
Information about the onboard Jetson
Last updated
The F-11 houses an Nvidia Orin NX 16GB module that delivers up to 100 TOPS for high-performance AI/ML applications.
This module sits on a Connect Tech Hadron carrier board with 500GB or 1TB SSD.
The onboard GPU is a 1024-core Ampere architecture GPU, boasting 32 tensor cores, as well as two NVDLA v2.0, an 8-core Arm Cortex-A78AE CPU, and 16GB 128-bit LPDDR5 memory.
More information is available in the Nvidia Data Sheet.
The currently supported Nvidia JetPack version is 5.1.2, which includes a special board support package (BSP) for the Hadron carrier board. This version comes with Ubuntu 20.04, and is equipped with Gstreamer 1.16.3. Most Nvidia frameworks and libraries can be preloaded onto the drone on request by contacting Jason, including CUDA and cuDNN.
More Information can be found on the Nvidia JetPack page.
Currently supported payloads include the Gremsy Vio (both G1 and F1), Sony LR1, Gremsy Pixy LR, GoPro Hero 12 Black, and Raspberry Pi camera V2. More payload and sensor integrations are being developed, and any special requests will be evaluated for potential integrations.
Developers may choose to integrate any payload or hardware that they choose. We are actively implementing a framework for end-to-end payload integration development that will allow users to integrate any payload with ease while allowing full control.
The Jetson is connected to the flight controller, and therefore the mavlink system, via UART. The baudrate that the this connection expects is 115200. We use mavlink router to communicate with the flight controller. Mavlink router takes over this serial port (/dev/ttyTCU0), and exposes a UDP server at 127.0.0.1:50051. The configuration is stored in /etc/mavlink-router/main.conf, which can be modified to add more endpoints. More information can be found here.
For development, MAVSDK and pymavlink can be used to integrate with the mavlink system. Note that MAVSDK is designed for PX4, whereas our system is built on top of Ardupilot, which may result in some functionality and mavlink messages being slightly different. Pymavlink is designed with Ardupilot in mind.
| Feature | Description |
|---|---|
AI Performance
100 TOPS
GPU
1024-core NVIDIA Ampere architecture GPU with 32 Tensor Cores
GPU Max Frequency
918MHz
CPU
8-core Arm® Cortex®-A78AE v8.2 64-bit CPU 2MB L2 + 4MB L3
CPU Max Frequency
2 GHz
DL Accelerator
1x NVDLA v2
DLA Max Frequency
614MHz
Vision Accelerator
1x PVA v2
Memory
16GB 128-bit LPDDR5 102.4GB/s
Video Encode
1x 4K60 (H.265) 3x 4K30 (H.265) 6x 1080p60 (H.265) 12x 1080p30 (H.265)
Video Decode
1x 8K30 (H.265) 2x 4K60 (H.265) 4x 4K30 (H.265) 9x 1080p60 (H.265) 18x 1080p30 (H.265)
CSI Camera
Up to 4 cameras (8 via virtual channels***) 8 lanes MIPI CSI-2 D-PHY 2.1 (up to 20Gbps)
PCIe
1 x4 + 3 x1 (PCIe Gen4, Root Port, & Endpoint)
USB
3x USB 3.2 Gen2 (10 Gbps) 3x USB 2.0
Networking
1x GbE
Other I/O
3x UART, 2x SPI, 2x I2S, 4x I2C, 1x CAN, DMIC & DSPK, PWM, GPIOs
Power
10W-25W
Mechanical
69.6mm x 45mm 260-pin SO-DIMM connector
