Repository files navigation

ELSA Robot

• To install teleop-twist-keyboard

sudo apt-get install ros-melodic-teleop-twist-keyboard

• To install gmapping clone and buildthis
• To install robot_localization, clone and buildthis

• The project has two packageselsa andros_arduino_bridge
• Follow the below instructions to build the project

cd ~/catkin_ws`source devel/setup.bashcatkin_make --only-pkg-with-deps elsacatkin_make --only-pkg-with-deps ros_arduino_bridge

• Follow the below commands to execute the project

roslaunch elsa elsabot.launch

• Use the below command to do keyboard teleop

rosrun teleop_twist_keyboard teleop_twist_keyboard.py

• To launch gmapping

roslaunch gmapping slam_mapping_pr2.lanch

• To save the map

rosrun map_server map_saver -f /tmp/my_map

• To use robot_localization package for filtered odometry data
  • roslaunch robot_localization ekf_template.launch --> This will be merged to the elsa project, as we are required to edit the EKF parameters according to our requirement

• Coordinate Transformations and Robot Dimensions can be found atdocs/Robot dimensions and coordinate transformations.pdf
• More info about the references used is provided in the reference section

• Publish imu to/imu/raw_data topic
• Publish encoder ticks to/ticks_pub/lwheel_ticks and/ticks_pub/lwheel_ticks topics
• Publish Odometry data to/odom topic
• Configured the ROS_Arduino_Bridge
• Debug the ROS_Arduino_Bridge
  • Solder Arduino Nano
  • Connect the Arduino Nano to the Motor_Driver
  • Edit the scripts for the second USB port (/dev/ttyUSB*)
• Publish encoder ticks to/arduino/lwheel_ticks and/arduino/lwheel_ticks topics
• Publish velocoties to/cmd_vel topic
  • Publish sample Twist msg -- Seems like there's an error/lag
  • Debug the motor_driving
• Finish the base_controller to subscribe to/cmd_vel topic to move the robot_base
• Write a node to Publish raw images to a ROS topic
  • Resolve cv_bridge and opencv4 issue for ROS - the issue is specific to ROS-melodic and OpenCV4 combination
• Transport Pi camaera v2 video stream using sensor_msgs/Image to a ROS topic
• Setup either YDLidar X4 for 2D map creation - needed extra cable to power the LIDAR, since the RPi cannotprovide enough power to drive the LIDAR motor
• visualized the tf and odometry in rviz
  • Fix the odometry axis and tf axis alignment - made sure all the axis are aligning in same direction
• rename the coordinate frames of the ydlidar to match with our convention
• fix encoder direction
• fix the motor direction
• Use teleop_twist_keyboard to move the robot -- works perfectly
• Use robot_localization package to filter the Odometry msg using ekf_lozalisation node
  • check if everything is working as expected -- need to tweak covariance matrix for better estimation
• Fix the direction issue with rviz visualization -- Problem is with Odometry calculation -- robot_localization is solving this
• Use slam techniques for mapping - Works
• Tune the PID controller

• Use camera data to obtain the images to do SFM and build the 3D_point_cloud_map

ELSA Result

• The below image shows the/base_link and/odom coordinate frames and trasformations between them in rviz

Coordinate Transformations (TF) in rviz

• The below image shows the robot movement in rviz based on the odometry published to/odom topic

Odometry of the ELSA Robot in rviz

• The below image shows the laserscan from YDLIDAR X2 being pubilshed to/scan

Laser Scan in rviz

• The below image shows the TF tree based on coordinate transforms being published to/tf

TF Tree

• The TF tree can be found atelsa/docs/frames.pdf
• The project uses theROSArduinoBridge Package. However, many modifications are made according to the requirements
• The Project also uses theSparkFun_RedBot_Arduino_Library Arduino Library. This library has been modified a little bit according to our requirement. The updated library can be found in/elsa/elsa/arduino/
• The Projects also uses few insights from the packagediff_drive
• The project refers several tutorials fromROS wiki
• The topics are published according to the coordinates frames as perREP103 standards andREP105 standards
• The YDLIDAR uses the package provided by EAI YDLIDARgithub
• The Project uses robobt_localization package to fuse imu and encoder-based Odometry data, to provide filtered pose. Technical documentation for the robot_localization package can be foundhere