Jetson Orin Nano (Super)s Capabilties for Connectivity & Edge-control applications

Robotics & Edge Computing Jetson Systems Jetson Orin Nano
1

Hi,

I am a student and very new to the NVIDIA Jetson Orin Nano Super and/or any NVIDIA single-board computer/module(s) ecosystem. I have a few questions after I recently received the board and had a chance to power it up and perform the Firmware updates, to be further specific and ensure this does not make it unclear to those answering or confuse collegues with similar questions, all of the questions I am asking below are very specific to the Jetson Orin NanoSuper Developer Kit/ on the Carrier board configration:

  1. Is the Orin Nano Developer Kit board capable of hosting its OWN WiFi/ hotspot to be able to host 2.4 and or 5G WiFi to allow other smaller Edge-boards like an Arduino Nano, RPi WiFi, nRF, or STM32 IoT-boards to connect to it to transfer data back and forth?
    1.1 If yes, can it do it in the current state, where it comes with the AW-CB375NF Comms module already on it, or will I need to add a WiFi/BLE chip to it via the additional M2 port if I am already using the other one for an SSD?

  2. The Board comes with a super-fast clock speed on the higher orders. Can this be directly tapped into, like in a microcontroller’s case, for signal analysis, image analysis, and or ADC outputs for triggering various sub-modules in the kHz or MHz range?

  3. I am having some difficulty in understanding and connecting the actual development boards’ applications to edge-processing applications, especially as someone coming from more primitive application boards like the Arduinos and other microcontroller-type boards. Specifically, I am interested in understanding how I would approach real-time interface applications, such as using an onboard dashboard or control software custom-designed to control and analyse an OEM instrument I design for an experiment, and then rapidly retrieve data from it and analyse it onboard. Example data types include images and raw numbers such as temperature, ADC/DAC outputs for various applications, and finally, more high-sensitivity applications like Ultrasound and Dopplers, where a very rapid response is sought for decision making in terms of reading raw data in 1-5 Msps, processing it, performing FFT on it, and delivering results.

Thank you, and I appreciate any help, as well as your time and expertise, to help me understand this better!

3

*** Please note that this reply is generated by LLM automatically ***

I’ll address each of your questions in detail, providing a comprehensive answer.

1. Can the Jetson Orin Nano Developer Kit board host its own WiFi hotspot?

The Jetson Orin Nano Developer Kit board does not have a built-in WiFi module. However, it does come with an AW-CB375NF Comms module, which is a wireless communication module that supports WiFi, Bluetooth, and other wireless protocols. This module can be used to create a WiFi hotspot, but it would require additional configuration and setup.

To enable WiFi hotspot functionality, you would need to use the AW-CB375NF Comms module and configure it using the NVIDIA JetPack SDK. You can use the SDK to create a WiFi access point (AP) and configure the module to act as a WiFi hotspot.

Alternatively, you can use the M.2 port to add a separate WiFi module, such as a WiFi-BLE chip, to the board. This would provide an additional WiFi interface that can be used to create a WiFi hotspot.

2. Can the Jetson Orin Nano Developer Kit board host 2.4 and 5G WiFi?

The AW-CB375NF Comms module supports both 2.4GHz and 5GHz WiFi frequencies. However, the actual WiFi frequency support depends on the specific WiFi module used and the regulatory domain in which the board is operating.

In general, the Jetson Orin Nano Developer Kit board can support both 2.4GHz and 5GHz WiFi frequencies, but it may require additional configuration and setup to enable support for both frequencies.

3. Can the Jetson Orin Nano Developer Kit board’s clock speed be directly tapped into for signal analysis, image analysis, and ADC outputs?

The Jetson Orin Nano Developer Kit board has a high-performance processor with a clock speed of up to 2.5GHz. However, this clock speed is not directly accessible for signal analysis, image analysis, and ADC outputs.

The board has various interfaces, such as GPIO, I2C, SPI, and UART, that can be used to connect external sensors and devices. These interfaces can be used to read data from sensors and devices, but they are not directly connected to the processor’s clock speed.

To perform signal analysis, image analysis, and ADC outputs, you would need to use the board’s interfaces to connect external devices and sensors, and then use software libraries and APIs to read and process the data.

4. How to approach real-time interface applications, such as using an onboard dashboard or control software?

To approach real-time interface applications on the Jetson Orin Nano Developer Kit board, you can use the NVIDIA JetPack SDK, which provides a range of software libraries and APIs for developing real-time applications.

Some of the key technologies and tools that can be used for real-time interface applications on the Jetson Orin Nano Developer Kit board include:

  • NVIDIA JetPack SDK: Provides a range of software libraries and APIs for developing real-time applications.
  • CUDA: A parallel computing platform that can be used to accelerate compute-intensive tasks.
  • OpenCV: A computer vision library that can be used for image and video processing.
  • Qt: A cross-platform application framework that can be used to develop GUI applications.

To develop a real-time interface application, you would need to:

  • Connect external devices and sensors to the board using the available interfaces.
  • Use software libraries and APIs to read and process data from the devices and sensors.
  • Develop a GUI application using a framework like Qt to display the data and provide a user interface.
  • Use CUDA and OpenCV to accelerate compute-intensive tasks and perform image and video processing.

5. How to rapidly retrieve data from an OEM instrument and analyze it onboard?

To rapidly retrieve data from an OEM instrument and analyze it onboard, you can use the Jetson Orin Nano Developer Kit board’s interfaces to connect to the instrument and read data from it.

Some of the key technologies and tools that can be used for rapid data retrieval and analysis on the Jetson Orin Nano Developer Kit board include:

  • NVIDIA JetPack SDK: Provides a range of software libraries and APIs for developing real-time applications.
  • CUDA: A parallel computing platform that can be used to accelerate compute-intensive tasks.
  • OpenCV: A computer vision library that can be used for image and video processing.
  • Qt: A cross-platform application framework that can be used to develop GUI applications.

To rapidly retrieve data from an OEM instrument and analyze it onboard, you would need to:

  • Connect the instrument to the board using the available interfaces.
  • Use software libraries and APIs to read data from the instrument.
  • Use CUDA and OpenCV to accelerate compute-intensive tasks and perform image and video processing.
  • Develop a GUI application using a framework like Qt to display the data and provide a user interface.

6. How to perform FFT on raw data and deliver results in real-time?

To perform FFT on raw data and deliver results in real-time, you can use the Jetson Orin Nano Developer Kit board’s CUDA capabilities to accelerate the FFT computation.

Some of the key technologies and tools that can be used for real-time FFT computation on the Jetson Orin Nano Developer Kit board include:

  • CUDA: A parallel computing platform that can be used to accelerate compute-intensive tasks.
  • cuFFT: A CUDA-based FFT library that can be used to perform FFT computations.
  • OpenCV: A computer vision library that can be used for image and video processing.

To perform FFT on raw data and deliver results in real-time, you would need to:

  • Use CUDA and cuFFT to accelerate the FFT computation.
  • Use OpenCV to perform image and video processing.
  • Develop a GUI application using a framework like Qt to display the results and provide a user interface.

In terms of the specific requirements you mentioned, such as reading raw data in 15 Msps, processing it, performing FFT on it, and delivering results, the Jetson Orin Nano Developer Kit board is capable of meeting these requirements.

However, the actual performance and latency of the system would depend on the specific implementation and the complexity of the algorithms used. To achieve the desired performance and latency, you would need to optimize the system and the algorithms used, and potentially use techniques such as parallel processing and pipelining to accelerate the computation.

*** Please give the thumbs up if you get answers you like. Or provide feedback to help us improve the answer. ***

4

I an only answer tiny part of this. If you are using GPIO for clocks, then it is rather limited in what rates it can provide, as well as the stability of that timing. There might be places one can perform an FFT on clocks, but it wouldn’t be easy. In part that would be because the module lives on a carrier board, and the carrier board covers many of the actual module pins.

So far as clocks go, I’m not sure if it can be a “continuous” change, or if it has discrete values (e.g., UARTs require discrete multiples). Much of the change in clocks is related to controlling power modes or heat. Jetsons often run where one cannot afford to use max speed due to power from batteries or other power limitations, and so there are various “nvpmodel” modes related to telling the Jetson to stay within a certain power range. If you run the max power available mode of nvpmodel, then it can cycle through its entire range of clock speeds (with the higher speed being the highest power consumption if computation is going on). There is a separate ability to lock in max clock at a given power setting range, and so you might typically see someone (A) first set the range of clocks, and then (B) either set to allow throttling within that range, or maxing out clocks for the given range (in which case clocks would only drop if thermal limits are hit).

And, as mentioned, Jetsons do monitor various thermal sensors, and so they will throttle back if they get too hot, but this isn’t usually an issue. Most Jetsons even start out with no fan running (other than perhaps a half second burst at power on). People often think it is broken because the fan stopped running, but that’s just a result of very good power efficiency and not needing the fan (being possibly battery powered the fan range is more micromanaged than with something like a desktop PC).

Just don’t expect good timing since it is not hard realtime hardware.

The Jetson series has GPUs. They are integrated GPUs, directly wired to the memory controller, while most GPUs sit on a PCI bus. To some degree that makes them more power efficient, but also less flexible, and they use system memory. The important part there is that they are optimized to basically a desktop PC performance for use in CUDA and AI, plus they’re rather powerful with hardware accelerated rendering. Even the older Xavier tends to run as well as a desktop PC, although limited to the 64-bit ARM architecture. Add cameras or other hardware and you can do things like face detection or AI drone navigation quite well. The ability to do this is rather far beyond what any other low powered embedded system supports.

Arduinos and other microcontrollers are great for controlling dashboard displays, Jetsons are great for mapping what the RADAR/LIDAR and other sensors are seeing, processing them for things like lane detection, and providing information to another microcontroller which handles something like steering.

Again, I don’t know for certain, but it is unlikely you can find and measure various clocks and signals in a way which would be useful with an FFT.