search

linorobot / linorobot2

Autonomous mobile robots (2WD, 4WD, Mecanum Drive)
Apache License 2.0
492 stars 165 forks source link
2wd 4wd amr autonomous diy diy-robot gazebo linorobot mecanum-wheel robotics robots ros ros2

readme

linorobot2

linorobot2

linorobot2 is a ROS2 port of the linorobot package. If you're planning to build your own custom ROS2 robot (2WD, 4WD, Mecanum Drive) using accessible parts, then this package is for you. This repository contains launch files to easily integrate your DIY robot with Nav2 and a simulation pipeline to run and verify your experiments on a virtual robot in Gazebo.

Once the robot's URDF has been configured in linorobot2_description package, users can easily switch between booting up the physical robot and spawning the virtual robot in Gazebo.

linorobot2_architecture

Assuming you're using one of the tested sensors, linorobot2 automatically launches the necessary hardware drivers, with the topics being conveniently matched with the topics available in Gazebo. This allows users to define parameters for high level applications (ie. Nav2 SlamToolbox, AMCL) that are common to both virtual and physical robots.

The image below summarizes the topics available after running bringup.launch.py. linorobot2_microcontroller

An in-depth tutorial on how to build the robot is available in linorobot2_hardware.

Installation

This package requires ros-foxy or ros-galactic. If you haven't installed ROS2 yet, you can use this installer script that has been tested to work on x86 and ARM based dev boards ie. Raspberry Pi4/Nvidia Jetson Series.

1. Robot Computer - linorobot2 Package

The easiest way to install this package on the robot computer is to run the bash script found in this package's root directory. It will install all the dependencies, set the ENV variables for the robot base and sensors, and create a linorobot2_ws (robot_computer_ws) on the robot computer's $HOME directory. If you're using a ZED camera with a Jetson Nano, you must create a custom Ubuntu 20.04 image for CUDA and the GPU driver to work. Here's a quick guide on how to create a custom image for Jetson Nano.

source /opt/ros/<ros_distro>/setup.bash
cd /tmp
wget https://raw.githubusercontent.com/linorobot/linorobot2/${ROS_DISTRO}/install_linorobot2.bash
bash install_linorobot2.bash <robot_type> <laser_sensor> <depth_sensor>
source ~/.bashrc

robot_type:

laser_sensor:

Sensors marked with an asterisk are depth sensors. If a depth sensor is used as a laser sensor, the launch files will run depthimage_to_laserscan to convert the depth sensor's depth image to laser scans.

depth_sensor:

Alternatively, follow this guide to do the installation manually.

2. Host Machine / Development Computer - Gazebo Simulation (Optional)

This step is only required if you plan to use Gazebo later. This comes in handy if you want to fine-tune parameters (ie. SLAM Toolbox, AMCL, Nav2) or test your applications on a virtual robot.

2.1 Install linorobot2 Package

Install linorobot2 package on the host machine:

cd <host_machine_ws>
git clone -b $ROS_DISTRO https://github.com/linorobot/linorobot2 src/linorobot2
rosdep update && rosdep install --from-path src --ignore-src -y --skip-keys microxrcedds_agent --skip-keys micro_ros_agent
colcon build
source install/setup.bash

2.2 Define Robot Type

Set LINOROBOT2_BASE env variable to the type of robot base used. Available env variables are 2wd, 4wd, and mecanum. For example:

echo "export LINOROBOT2_BASE=2wd" >> ~/.bashrc
source ~/.bashrc

You can skip the next step (Host Machine - RVIZ Configurations) since this package already contains the same RVIZ configurations to visualize the robot.

3. Host Machine - RVIZ Configurations

Install linorobot2_viz package to visualize the robot remotely specifically when creating a map or initializing/sending goal poses to the robot. The package has been separated to minimize the installation required if you're not using the simulation tools on the host machine.

cd <host_machine_ws>
git clone https://github.com/linorobot/linorobot2_viz src/linorobot2_viz
rosdep update && rosdep install --from-path src --ignore-src -y 
colcon build
source install/setup.bash

Hardware and Robot Firmware

All the hardware documentation and robot microcontroller's firmware can be found here.

URDF

1. Define robot properties

linorobot2_description package has parameterized xacro files that can help you kickstart writing the robot's URDF. Open .properties.urdf.xacro in linorobot2_description/urdf directory and change the values according to the robot's specification/dimensions. All pose definitions must be measured from the base_link (center of base) and wheel positions (ie wheel_pos_x) are referring to wheel 1.

For custom URDFs, you can change the urdf_path in description.launch.py found in linorobot2_description/launch directory.

Robot Orientation:

--------------FRONT--------------

WHEEL1 WHEEL2 (2WD/4WD)

WHEEL3 WHEEL4 (4WD)

--------------BACK--------------

Build the robot computer's workspace to load the new URDF:

cd <robot_computer_ws>
colcon build

The same changes must be made on the host machine's .properties.urdf.xacro if you're simulating the robot in Gazebo. Remember to also build the host machine's workspace after editing the xacro file.

cd <host_machine_ws>
colcon build

2. Visualize the newly created URDF

2.1 Publish the URDF from the robot computer:

ros2 launch linorobot2_description description.launch.py

Optional parameters for simulation on host machine:

2.2 Visualize the robot from the host machine:

The rviz argument on description.launch.py won't work on headless setup but you can visualize the robot remotely from the host machine:

ros2 launch linorobot2_viz robot_model.launch.py

Quickstart

All commands below are to be run on the robot computer unless you're running a simulation or rviz2 to visualize the robot remotely from the host machine. SLAM and Navigation launch files are the same for both real and simulated robots in Gazebo.

1. Booting up the robot

1.1a Using a real robot:

ros2 launch linorobot2_bringup bringup.launch.py

Optional parameters:

Always wait for the microROS agent to be connected before running any application (ie. creating a map or autonomous navigation). Once connected, the agent will print:

| Root.cpp             | create_client     | create
| SessionManager.hpp   | establish_session | session established

The agent needs a few seconds to get reconnected (less than 30 seconds). Unplug and plug back in the microcontroller if it takes longer than usual.

1.1b Using Gazebo:

ros2 launch linorobot2_gazebo gazebo.launch.py

linorobot2_bringup.launch.py or gazebo.launch.py must always be run on a separate terminal before creating a map or robot navigation when working on a real robot or gazebo simulation respectively.

2. Controlling the robot

2.1 Keyboard Teleop

Run teleop_twist_keyboard to control the robot using your keyboard:

ros2 run teleop_twist_keyboard teleop_twist_keyboard

Press:

2.2 Joystick

Pass joy argument to the launch file and set it to true to enable the joystick. For example:

ros2 launch linorobot2_bringup bringup.launch.py joy:=true

Press Button/Move Joystick:

3. Creating a map

3.1 Run SLAM Toolbox:

ros2 launch linorobot2_navigation slam.launch.py

Optional parameters for simulation on host machine:

For example:

ros2 launch linorobot2_navigation slam.launch.py rviz:=true sim:=true

3.1 Run rviz2 to visualize the robot from host machine:

The rviz argument on slam.launch.py won't work on headless setup but you can visualize the robot remotely from the host machine:

ros2 launch linorobot2_viz slam.launch.py

3.2 Move the robot to start mapping

Drive the robot manually until the robot has fully covered its area of operation. Alternatively, you can use the 2D Goal Pose tool in RVIZ to set an autonomous goal while mapping. More info here.

3.3 Save the map

cd linorobot2/linorobot2_navigation/maps
ros2 run nav2_map_server map_saver_cli -f <map_name> --ros-args -p save_map_timeout:=10000.

4. Autonomous Navigation

4.1 Load the map you created:

Open linorobot2/linorobot2_navigation/launch/navigation.launch.py and change MAP_NAME to the name of the newly created map. Build the robot computer's workspace once done:

cd <robot_computer_ws>
colcon build

Alternatively, map argument can be used when launching Nav2 (next step) to dynamically load map files. For example:

ros2 launch linorobot2_navigation navigation.launch.py map:=<path_to_map_file>/<map_name>.yaml

4.2 Run Nav2 package:

ros2 launch linorobot2_navigation navigation.launch.py

Optional parameter for loading maps:

Optional parameters for simulation on host machine:

4.3 Run rviz2 to visualize the robot from host machine:

The rviz argument for navigation.launch.py won't work on headless setup but you can visualize the robot remotely from the host machine:

ros2 launch linorobot2_viz navigation.launch.py

Check out Nav2 tutorial for more details on how to initialize and send goal pose.

navigation.launch.py will continue to throw this error Timed out waiting for transform from base_link to map to become available, tf error: Invalid frame ID "map" passed to canTransform argument target_frame - frame does not exist until the robot's pose has been initialized.

Troubleshooting Guide

1. The changes I made on a file are not taking effect on the package configuration/robot's behavior.

2. [slam_toolbox]: Message Filter dropping message: frame 'laser'

3. target_frame - frame does not exist

4. Weird microROS agent behavior after updating the Linux/ROS

Useful Resources:

https://navigation.ros.org/setup_guides/index.html

http://gazebosim.org/tutorials/?tut=ros2_overview