A ROS project developed as part of ME495 - Embedded Systems in Robotics course at Northwestern University (NU).

To launch (in a ROS workspace), type in terminal:

cd ~/baxter_ws/
. baxter.sh
rosrun baxter_tools enable_robot.py -e
roslaunch baxter_two baxter_two.launch

Use baxter to pick and place objects from one location to another

The following sections describe contents of this package and how to use them:

The following packages need to be installed for baxter_two package to work:

1. Baxter SDK - to be installed in baxter_ws/src directory
2. MoveIt!
3. moveit_python by Michael Ferguson - to be installed in baxter_ws/src directory
4. Baxter Simulator - v1.1 (ONLY needed if NOT working with actual baxter robot)

The baxter_two package consists of following main files:

• baxter_two.launch: A launch file that starts the following nodes:

  • joint_action_trajectory_server.py (required for MoveIt!)
  • MoveIt! with rviz
  • baxter_img.cpp
  • baxter_pnp.py

• baxter_img.cpp: subscribes to the /cameras/right_hand_camera/image topic, calculates object locations using OpenCV tools, and publishes this information to baxter_pnp node via a PoseArray() message

• baxter_pnp.py: subscribes to the topic (called Dpos) published by baxter_img, and calls PlanningSceneInterface and MoveGroupInterface classes from moveit_python bindings to add collision objects and move baxter's arms respectively

  • Allows both arms to be moved at the same time
  • Allows velocity scaling using max_velocity_scaling_factor
  • Uses baxter_interface for gripper control

This section gives a step-by-step approach to run a successful baxter pick and place with this package.

1. Items needed - a table of height ~ 0.6 m; square blocks (with green top surface); a baxter robot with electric parallel grippers; an ethernet cable (to connect baxter to computer)
2. Set up networking with baxter
3. Install the required packages as outlined in Prerequisites section
4. Clone this package to your baxter_ws/src directory
5. Source baxter.sh file, enable the robot, and launch baxter_two.launch as follows

cd ~/baxter_ws/
. baxter.sh
rosrun baxter_tools enable_robot.py -e
roslaunch baxter_two baxter_two.launch

The following rqt_graph shows a ROS computation graph for baxter pick and place highlighting all important nodes and topics during operation:

• Camera Calibration: We are looking into improving the camera calibration. Currently, variations in surrounding lighting affect the object detection. Using more finely-tuned methods of calibrating the camera will minimize the lighting issue.

• GUI Development: We are also looking to create a Graphical User Interface (GUI) that allows the user to click an object in the virtual world (e.g in Gazebo), and pick and place that particular object. This GUI is currently under development.

• Abhishek Patil
• Alan Hong
• Hanlin Wang
• Vince Martinez