This is task defined project to asses the new hirings for Robotics Software Engineer It helps the company to evaluate the applicants knowledge and skills in the tools and frameworks used in the department.
- Implementation and coding skills
- C++ and Python profcincy
- Robot Operation Systems (ROS)
- Robotics Fundementals
- Autonomous Navigation Fundementals
- GUI development
- Software Integration
- Fork the repo to your account, and reply to the email with your repo fork that contains the soloutions once you finish (Any reply after two weeks of the email wil not be accepted!!!).
- Try to utilize known/open-source tools as possible.
- Edit the README.md file in your fork, and add the steps and exxplination of your solution for each milestone.
You are given a ROS1 workspace that contains a ready to use setup for a car-like robot equibbed with a depth camera and a 3D LiDAR, placed in a virtual world within Gazebo Simulator. The target goal for this project, is to develop a minimal user controlled autonomous navigation functionality for the robot provided.
- Add the models used in the world file:
export GAZEBO_MODEL_PATH=$GAZEBO_MODEL_PATH:{Add your workspace path}/src/mybot_pkg/models- Check and install the corresponding dependencies:
# Available in RSE_ws
./install.sh- Launch the robot inside Gazebo:
# Launch the Robot inside Gazebo World
roslaunch mybot_pkg gazebo.launchTo achieve the desired goal, we break it down to smaller to be achieved in order based on its debendency for the the next step.
First, you need to map the robot world so it can understand it for later operations.
I have used rtabmap_ros for mapping.
You need to use gazebo_mapping.launch from mybot_pkg for bringing up the robot.
Then use rtab_mapping.launch from mapping package to generate the map.
Then save the map using map_saver.launch from mapping package.
Next, to move the robot autonomously around the map you need to localize the robot in real-time without using SLAM (offline localization).
I have used pointcloud_to_laserscan package, then use it for offline localisation.
Once you have a represntation of the environment and you can localize your robot within it, you can then start the autonomous navigation of the robot.
I have used move_base and amcl from navigation stack to autonomously navigate using the saved map with offline navigation .
You need to use gazebo_navigation.launch from localisation_and_navigation package
Then use rtab_navigation.launch from localization_and_navigation package
Then use rviz.launch from mybot_pkg package to send goals.
To make sure a smother operation of the robot when deploying it in the field, it's better to have a user friendly remote operation capabilties.
Develop a GUI that allows use to remotly control the robot from outside a ROS environment.
Feel free to pick whatever framework or stack you like (C++ Qt, Python, Web-based GUI, etc...).
NOTE: Implement only the basic functionality (Drive, Steer).
For this task you can use react_ros_web_gui module, to setup , install nvm, then use npm i, inside the module to install the module
Run websocket.launch from rosbridge_server package
Then use npm start inside the react_ros_web_gui to run the gui. Here you can use the joystick to move the robot around.
### 5 - User Defined Navigation (Open) Develop the previous milestone and adopt it so the user of your GUI can also perform the Navigation functionality (Sendg Waypoints/Goal, Mapping, etc...).
you can use the same web_gui to type in goal coordinates you want the robot to navigate to, or use rviz.launch from mybot_pkg and use 2DNavgoal to start navigation. ### (Optional) - Develop an Odometry Source for the robot The very first required components to start working on any autonomous functionality are the position, orientation, velocity feedback of the robot.
If we ignore the Odometry feedback provided by Gazebo, based on the robot description provided and the sensor set, develop a node that produce an Odometry feedback as accurate as possible.
GOOD LUCK! Collapse
has context menu