Sphere Follower Robot

Theodor Amariucai

I started the project with Sonia Marshall as part of our coursework for INFR09019 Introduction to Vision and Robotics. I then kept working on the project until I brought it from the state documented in folder 'Initial Assignment' up to the present state.

A very useful resource which helped me understand the Denavit-Hartenberg parameters was the Robotic Arm Kinematic GUI (part of MRPT). I used it to check my calculations which can be found here.

Before anything, make sure you:

source devel/setup.sh

Launch controller straight away:

roslaunch sphere-follower-robot control.launch

Launch the basic project:

roslaunch sphere-follower-robot spawn.launch

Run code on the basic project:

Activate temporary python environment with Python 2.7 needed for ROS

source temp-python/bin/activate

For basic functionality please run:

rosrun sphere-follower-robot BlobsEstimator.py rosrun sphere-follower-robot TargetEstimator.py rosrun sphere-follower-robot JointAnglesEstimator.py

To move the robot, please run:

rosrun sphere-follower-robot Controller.py

To get the results of forward kinematics, please run:

rosrun sphere-follower-robot ForwardKinematics.py

Manually move the robot:

rostopic pub -1 /robot/joint1_position_controller/command std_msgs/Float64 “data: 1.0”

Produce graphs for the target estimates:

For each one, run the command, click on the arrow button to set axes (choose around 50 seconds) and save as an image.

• rqt_plot /target_position_estimate/data[0] /target/x_position_controller/command/data
• rqt_plot /target_position_estimate/data[1] /target/y_position_controller/command/data
• rqt_plot /target_position_estimate/data[2] /target/z_position_controller/command/data

Produce graphs for control accuracy:

• rqt_plot /target_position_estimate/data[0] /blobs_pos/data[9]
• rqt_plot /target_position_estimate/data[1] /blobs_pos/data[10]
• rqt_plot /target_position_estimate/data[2] /blobs_pos/data[11]

Sphere Follower Robot Demo, captured in Gazebo