2021-08-28 Jatson Nano

Jatson Nano 

---

Introduction

With the electronics and software framework at a good point, next step for robotic development is to experiment with some AI. Three options were considered:

• Raspberry Pi 3B+ 
• Raspberry Pi 3B+ with Intel Movidius
• Jatson Nano 2GB

Nvidia produced an OS image to integrate their CUDA cores, the Jatson Nano board and ROS into an AI robot. The choice was made to obtain a Jetson Nano 2GB

The scope of this post is to power up the Jetson Nano and test the basic configuration procedure, compatibility with the camera module and the TPS55386 voltage regulators developed for my robots.

---

 Setup

 Use the TPS55386 voltage regulator designed for the Raspberry Pi based robot to power up the Jatson nano through the GPIO using an adapter. Connect the Raspicam V1.3. Add a USB Hub. Add an HDMI to VGA adapter.

 

---

 OS Configuration

• Download Nvidia OS image for Jatson Nano 2GB
• Power up and check screen
• Connect Keyboard and Mouse
• Complete configuration
• Launch Demo and check stability
  

---

Camera Test

The Raspicam V1.3 doesn't show on /dev/video0. After some research, Nvidia only support the Raspicam V2.0 IMX219 sensor on the MIPI CSI connector.

A regular USB camera shows up no problem because it uses the default linux V4L2 driver. Using gstreamer the camera streams correctly to the screen.

---

Conclusions

The voltage regulator is able to power up the Jatson Nano 2GB at full tilt no problem.

The Raspicam support is bad. I need to buy Raspicam 2.0 Modules.

It is disappointing that it's not easy to do an headless setup and no embedded wifi.

---