IKFast for Universal Robots

IKFast libraries for UR3, UR5, and UR10 both the CB and e-series.

Inspired by https://github.com/andyzeng/ikfastpy

Forward/Inverse Kinematics

This library computes the forward and inverse for Universal Robots from the base_link frame to the tool0 frame. This only applies to the ideal robot model defined by their URDF. The real robots have small differences in each joint that are not considered here, so expect to see position/orientation errors from submillimeter errors to a few centimeters.

Installation

Easy to install and use:

1. install dependencies

sudo apt-get install libblas-dev liblapack-dev
pip install --user numpy Cython

3. install using pip
   

git clone https://github.com/cambel/ur_ikfast.git
cd ur_ikfast
pip install -e .

It takes a few minutes to compile the IKfast libraries.

Example

from ur_ikfast import ur_kinematics

ur3e_arm = ur_kinematics.URKinematics('ur3e')

joint_angles = [-3.1, -1.6, 1.6, -1.6, -1.6, 0.]  # in radians
print("joint angles", joint_angles)

pose_quat = ur3e_arm.forward(joint_angles)
pose_matrix = ur3e_arm.forward(joint_angles, 'matrix')

print("forward() quaternion \n", pose_quat)
print("forward() matrix \n", pose_matrix)

# print("inverse() all", ur3e_arm.inverse(pose_quat, True))
print("inverse() one from quat", ur3e_arm.inverse(pose_quat, False, q_guess=joint_angles))

print("inverse() one from matrix", ur3e_arm.inverse(pose_matrix, False, q_guess=joint_angles))

For a new robot just create the ikfast database (.cpp) following one of these tutorials:

• http://docs.ros.org/kinetic/api/framefab_irb6600_support/html/doc/ikfast_tutorial.html
• http://docs.ros.org/kinetic/api/moveit_tutorials/html/doc/ikfast/ikfast_tutorial.html