Getting Started with CUSub

This guide assumes you have an understanding of ROS build environments and git version controlling. Some Helpful resources helpful resources can be found at the end of this page.

Setting up your workspace

The CUSub software stack consists of 4 meta-packages that contain all the runtime code for running the sub in sim and on the platform.

CUSub Sim: Contains all of the code related to the simulated underwater environment and descriptions of the subs.

CUSub Common: Hosts all of the drivers, motor controllers, sensor fusion, debugging tools and basic waypoint navigation server.

cusub_perception: image classification, feature extraction, and localization of objects.

cusub_cortex: Top level logic for controlling the sub including the state machine, global mapper, and bringup scripts for running the entire system as a whole.

Note

There is also a Dockerfile install.

In order to setup the CUSub workspace clone the repo into the src folder of you catkin_ws

cd ~/<your_catkin_ws>/src
git clone https://github.com/CU-Robosub/cusub

The software stack depends on a few external repositories that have been included as submodules. To initialize them:

cd cusub
git submodule init
git submodule update

note that as the submodules are updated by the respective repo owners it may be necessary to re-update to the latest code to ensure compatibility.

Library Dependencies

In order to build the sim package the protobuf C compiler is required

sudo apt install protobuf-compiler protobuf-c-compiler

Building the system

The preferred method of building the CUSub software environment involves using the catkin build tool rather than the catkin_make tool. to install catkin build

sudo apt install python-catkin-tools

once the catkin build tools are installed you can build each meta-package individually to ensure your environment is properly setup. first move back to the root of your ws

cd ~/<your_caktin_ws>

The recommended order is:

catkin build cusub_sim

catkin build cusub_common

To get started working with the sub it is not necessary to build the perception or cortex meta-packages and it is recommended that you have these two packages working before adding the perception and cortex packages.

catkin build cusub_perception

catkin build cusub_cortex

Docker install

Warning

Not quite fully polished.

Still has some issues but the core commands to make sure everything is installed to get the sub running are there.

Nvidia Docker setup

Install Docker

# Update system
$ sudo apt-get update
$ sudo apt-get upgrade
# Docker Dependencies
$ sudo apt-get install apt-transport-https ca-certificates curl software-properties-common
# Add docker PGP key
$ curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo apt-key add -
# Add docker repo
$ sudo add-apt-repository "deb [arch=amd64] https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable"
# Install docker
$ sudo apt-get update
$ sudo apt-get install docker-ce
# Test docker
$ docker -v
# Docker version 19.03.12, build 48a66213fe
# add user to docker group so you dont need sudo when using it
$ sudo usermod -aG docker Username
# Log out and back in to apply changes, or run
$ su - Username

Install Nvidia-Docker

# Add Nvidia docker key and source list
$ curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add -$ curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list
# Update
$ sudo apt-get update
# Install docker and reload daemons
$ sudo apt-get install -y nvidia-docker2
$ sudo pkill -SIGHUP dockerd
# Make sure latest NVidia drivers are installed, replace number "450" with latest
$ sudo apt-get install nvidia-450
# Test nvidia docker
docker run --runtime=nvidia --rm nvidia/cuda:9.0-base nvidia-smi
# Ouput should be similar to:
Sun Aug  2 00:29:45 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 440.95.01    Driver Version: 440.95.01    CUDA Version: 10.2     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce RTX 2060    On   | 00000000:01:00.0  On |                  N/A |
| N/A   49C    P8     9W /  N/A |    386MiB /  5926MiB |      2%      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
+-----------------------------------------------------------------------------+

Note

May need to reboot for this to work depending on if kernel was updated

Download the Dockerfile.

docker build --tag=robosub:melodic-bionic ROS-Docker
# Need to clone latest sources
# Need to build in custom Gazebo 9.4
sudo apt install xpra
# Download x11docker and run
git clone https://github.com/mviereck/x11docker
./x11docker --runtime=nvidia --xpra --user=root --share=/dev/input/js0 robosub:melodic-bionic xterm
roslaunch teleop joy.launch start_joy:=true joy_device:=/dev/input/js0 setpoint:=false namespace:=leviathan/cusub_common

References: