ros_environment

ROS source code that for visual servo controller and localization

System: Ubuntu 16.04 LTS

ROS Version: Kinetic

Hardware: NVIDIA Jeston TX2

Software architecture

modules
- 0_driver
- 1_perception_cv
- 2_perception_others
- 3_estimator
- 4_planning
- 5_decision
- 6_control
tools

Most used

0_driver: CAN driver; dynamic reconfigure; rqt_multiplot.
1_perception_cv: RMComputerVision, the visual frontend.
3_estimator: wheel_odom, wheel_driver.
6_control: the visual servo control backend, handling control and target velocity estimation.

Get started

If you compile the source code for the first time, start with the minimum code necessary: The CAN driver, the 1_perception_cv, and 6_control.

install Robotics Operating System Kinetic in your Ubuntu computer http://wiki.ros.org/ROS/Installation
cd ~ && mkdir ws && cd ws
git clone --recursive git@github.com:robomasterhkust/ros_environment.git src
git submodule update
sudo cp 1_perception_cv/RMComputerVision/lib/libMVSDK.so /usr/lib
cd ~/ws && catkin_make

Toolchain

install the following dependency for fast development sudo apt-get install openocd minicom -y

sudo apt-get install terminator cmake vim htop libmuparser-dev ntp* -y

install these dependency for the visualization tool rqt_multiplot sudo apt-get install ros-kinetic-rqt libqwt-dev libqwt-qt5-dev ros-kinetic-rqt-multiplot -y

set up the .bashrc for speed up the development process

vim ~/.bashrc

and add the following:

alias cmk='cd ~/ws && catkin_make -j4 -l4 && source devel/setup.bash && cd -'

alias ll='ls -alF'

alias la='ls -A'

alias l='ls -cF'

source ~/ws/devel/setup.bash

and source ~/.bashrc after saving the vim file.

install Computer vision related dependences

openCV, Eigen

Reference I: RMOC open sourced repository https://github.com/robomasterhkust/RoboRTS

Reference II: My TA GAO Wenliang's repository for visual inertial localization

https://github.com/gaowenliang

Frame definition(right handed, all z axises heading upward)

World frame. At t = 0, x axis points to the starting direction of gimbal gunner, origin is at the starting rotation center of gimbal.
Chassis frame. Any time instant, x axis points to current direction of soldier heading, origin is at the current camera position on the chassis.
Transition frame. At any time instant, x axis points to current direction of soldier heading, origin is at the current rotation center of gimbal.
Gimbal frame. At any time instant, x axis points to current direction of gimbal gunner, origin is at the current rotation center of gimbal.

For NVIDIA TX2

flycapture has one extra step to configure, see https://www.ptgrey.com/tan/10699
for rqt_multiplot, sudo apt install libqwt-qt5-dev libqwt5-qt4 -y, and in the running sequence, use rosrun rqt_multiplot rqt_multiplot --force-discover

Network setting

-- rmsoldierX uses ip 192.168.1.22X in ASUS wifi network. All usernames are victory. -- copy the file tools/lsusb.sh to /etc

Attention

Never add CMakeList.txt to git repo.
Never add CMakeList.txt to git ignore.
There will be reminder that CMakeList.txt is not tracked in git. Just leave it there. Every computer has different src CMakeList.txt which linked to the place where ros is installed.

Multi-machine debugging

Two ros machines, one Intel Nuc, one TX2.

Follow the guide line here: https://askubuntu.com/questions/22835/how-to-network-two-ubuntu-computers-using-ethernet-without-a-router.

Set the ip address of nuc to be 10.0.0.2, tx2 to be 10.0.0.1. Nuc is the ros master.

export ROS_MASTER_URI=http://10.0.0.2:11311 on both computers.

To debug using another ROS machine, using multimachine by modifying the /etc/hosts and setting three environment variables in the ~/.bashrc

On both machines, add the ip address of the itself and the slave in the /etc/hosts, for instance, like this:

127.0.0.1       localhost
127.0.1.1       desktop
10.0.0.2   	rmsoldier#
10.0.0.1   	tegra-ubuntu

Then launch roscore in the master machine, and all nodes don't need to launch roscore and can receive topics from each others.

hkustenterprize/ros_environment