We developed
a software library that provides the dynamic model of a
KUKA Lightweight Robot IV (LWR IV) in two forms. The first one calculates the matrices and vectors
of the inverse dynamics model required by many model-based
motion controller. The second form computes a simulation
model (and the respective Jacobian matrix) to be used
by SUNDIALS, an implicit solver of ordinary differential equation.
The models are computed in pure C. Additionally, a Python interface is included for easy prototyping.
Consider citing our paper:
V. Bargsten, P. Zometa, and R. Findeisen.
"Modeling, parameter identification and model-based control of a robotic manipulator."
In: Control Applications (CCA), IEEE International Conference on.
pp. 134-139. IEEE. 2013.
Simulation of a trajectory tracking motion controller (also see video below).
Controller: feedback linearization.
Trajectory: a circle in a vertical plane.
Hardware: a computer based on a Intel Xeon processor.
Software: The controller was implemented in Python, using the dynamic model we developed.
SUNDIALS is used as implicit ODE solver. The 3-D simulation is based on Blender. All the development was made using free software.
Implementation of a model-based trajectory tracking controller. The motion controller linearizes the highly nonlinear dynamics via feedback linearization.
We additionally use a linear controller to stabilize the system.
Each joint behaves as a critically damped mass-spring-damper system,
which is clearly seen after the disturbances introduced by hand.
Controller: feedback linearization.
Trajectory: a circle in a vertical plane.
Hardware: KUKA LWR IV together with computer based on a Intel Xeon processor.
Software: The controller was implemented in Python, using the dynamic model we developed to linearize the system dynamics.