learn_kinematics

Forward and Inverse Kinematics of Baxter Robot

PLEASE ADD 0.01 TO Z COORDINATE BEFORE CALCULATING IK. SEE LINE 25 IN ik_test.py FOR MORE INFO

Dependencies

• moveit!
• baxter_pykdl
• trac_ik

Install trac_ik

1. Open terminal or press CTRL+ALT+T
2. Run following command

sudo apt-get install ros-${ROS_DISTRO}-trac-ik

For example, in the case of ROS Indigo, the following command is equivalent to the one above: sudo apt-get install ros-indigo-trac-ik

Installation

• Install baxter_pykdl package by downloading the package from git into the workspace src folder and running catkin_make.

• Install moveit dependecies: sudo apt-get install ros-indigo-moveit-ros

• Uncompress the downloaded trac_ik archive from the website

• Copy all four directories trac_ik, trac_ik_example, trac_ik_kinematics_plugin and trac_ik_lib into ros workspace

• Run catkin_make command from ros workspace

• If the abpve command shows following error- fatal error: nlopt.hpp: No such file or directory #include <nlopt.hpp> please run sudo apt-get install libnlopt-dev

• Run again catkin_make command

• Download or clone the repository

• Go to the ros workspace, i.e., cd ros_ws/

• Run catkin_make

• If the abpve command shows following error- fatal error: learn_kinematics/fast_kinematics_service.h: No such file or directory, please run catkin_make again

Programs

fkin.py, ikin.py, fkin2.py

These programs use the baxter_pykdl package which requires the robot urdf description to perform IK or FK.

Please connect to Baxter robot or run the baxter simulator for these programs to work properly.

fast_kinematics_pub.cpp

This program listen to rostopic fast_kinematics_listener for receiving end-effector pose from user. Later on it performs inverse kinematic analysis and publishs the result to another rostopic fast_kinematics_publisher

fast_kinematics_srv.cpp

This is a service implementaion for kinematic analysis. It creates a service fast_kinematics_service, which perfroms inverse kinematic analysis on request from client.

How to use

fast_kinematics_pub.cpp

• First, you need to publish std_msgs::Float64MultiArray containing end-effector pose (position in cartesian system and orientaion in quaternion system) to the rostopic fast_kinematics_listener as input.
• Now, run the subscriber and publisher by executing following command roslaunch learn_kinematics fast_kinematics_pub.launch
• The output joint angles can be seen by following command rostopic echo fast_kinematics_publisher

You may notice some error initially, as the subscriber haven't receieved any input and publisher starts publishing default values causing this error

fast_kinematics_srv.cpp

• Run the service by following command roslaunch learn_kinematics fast_kinematics_srv.launch
• Run the service client, which calls fast_kinematics_service with learn_kinematics::fast_kinematics_service service data type. Currently, it is done in c++ and python in following way -
• C++: Run the client by following command rosrun learn_kinematics fast_kinematics_srv_test
• Python: Run the client by following command rosrun learn_kinematics fast_kinematics_srv_test.py