构建属于你自己的Jetbot

1.     烧录Jetpack 镜像到SD卡  Jetson Download Center

        使用的账户和密码均为Jetbot

        确认使用自动登录

2.     将SD卡插入Jetson Nano 并开机进入命令行

3.     启用IC2权限

        sudo usermod -aG i2c $USER

4.     安装pip和python依赖项

        sudo apt-get update

        sudo apt install python3-pip python3-pil

        sudo pip3 install --upgrade numpy

5.     安装Tensorflow

        sudo apt-get update

        sudo apt-get install libhdf5-serial-dev hdf5-tools libhdf5-dev zlib1g-dev zip libjpeg8-dev liblapack-dev libblas-dev gfortran

        sudo apt-get install python3-pip

        sudo pip3 install -U pip testresources setuptools==49.6.0

        sudo pip3 install --pre --extra-index-url https://developer.download.nvidia.com/compute/redist/jp/v42 tensorflow-gpu

    注：验证Tensorflow安装在命令行输入python3进入python环境，然后在python环境内输入import tensorflow，如果没有报错，则tensorflow安装成功

6.     安装PyTorch

        注：需要先下载该文件并上传至/home/jetbot/目录   点我进入下载界面

        sudo pip3 install torch-1.0.0a0+18eef1d-cp36-cp36m-linux_aarch64.whl

        sudo pip3 install torchvision

7.     安装Traitlets

        sudo -H python3 -m pip install traitlets==4.3.3

8.     安装Jupter Lab依赖项

        sudo apt update

        sudo apt -y install curl dirmngr apt-transport-https lsb-release ca-certificates

        curl -sL https://deb.nodesource.com/setup_10.x | sudo bash

        sudo apt update

        sudo apt -y install nodejs npm

9.     安装Jupter Lab

        sudo pip3 install jupyter jupyterlab

        sudo jupyter labextension install @jupyter-widgets/jupyterlab-manager

        sudo jupyter labextension install @jupyterlab/statusbar

        jupyter lab --generate-config

        # 设置Jupyter Lab密码

        jupyter notebook password

        # 修改Jupyter Lab配置文件

        cd ~/jupyter

        sudo nano jupyter_notebook_config.py

        # 向配置文件末尾添加如下两行

        # c.NotebookApp.allow_origin = '*’

        # c.NotebookApp.ip = '0.0.0.0’

10.  安装这个Repo

        cd ~

        sudo apt install python3-smbus

        # 将Jetbot项目从Github克隆下来

        git clone https://github.com/NVIDIA-AI-IOT/jetbot

        cd jetbot

        sudo apt-get install cmake

        sudo python3 setup.py install

11.  安装Jetbot服务

        cd jetbot/utils

        python3 create_stats_service.py

        sudo mv jetbot_stats.service /etc/systemd/system/jetbot_stats.service

        sudo systemctl enable jetbot_stats

        sudo systemctl start jetbot_stats

        python3 create_jupyter_service.py

        sudo mv jetbot_jupyter.service /etc/systemd/system/jetbot_jupyter.service

        sudo systemctl enable jetbot_jupyter

        sudo systemctl start jetbot_jupyter

12.  创建文件交换（SWAP）

        sudo fallocate -l 4G /var/swapfile

        sudo chmod 600 /var/swapfile

        sudo mkswap /var/swapfile

        sudo swapon /var/swapfile

        sudo bash -c 'echo "/var/swapfile swap swap defaults 0 0" >> /etc/fstab'

13.  将Jetbot notebook复制到/home/jetbot/下

        cp -r ~/jetbot/notebooks ~/Notebooks

参考文章：

    https://github.com/NVIDIA-AI-IOT/jetbot/wiki/Create-SD-Card-Image-From-Scratch

    https://forums.developer.nvidia.com/t/official-tensorflow-for-jetson-nano/71770

    https://computingforgeeks.com/installing-node-js-10-lts-on-ubuntu-18-04-16-04-debian-9/