Robot Dynamics Identification

Citation Author(s):
Giacomo
Golluccio
University of Cassino and Southern Lazio
Giuseppe
Gillini
University of Cassino and Southern Lazio
Alessandro
Marino
University of Cassino and Southern Lazio
Gianluca
Antonelli
University of Cassino and Southern Lazio
Submitted by:
Laboratory Robotics
Last updated:
Wed, 06/10/2020 - 10:20
DOI:
10.21227/e5t7-pw09
Data Format:
License:
0
0 ratings - Please login to submit your rating.

Abstract 

This matlab code allows to reproduce some of the constrained and unconstrained dynamic identification techniques for open-chain robots. Below you can find an summary of the underlying research.

 

"Identification of dynamic parameters of robots is a long standing topic in robot control. Recently, the research witnessed a renewal of the activity due to some interesting results concerning issues involving the physical consistency of the obtained parameters leading to constraints, for instance, on  mass, first moment of mass and  inertia matrix of each rigid link of the robot.  This is motivated by  high  performance required by model-based control  of, for example, legged or surgical robots and to obtain realistic simulations of the same systems. The objective of this article is twofold. First, we review and experimentally compare on a Jaco 7 Degrees of Freedom anthropomorphic arm the latest results on the identification of the dynamic parameters which range from Least Mean Squares to constrained non-linear optimisation and Linear-Matrix-Inequality Semi-Definite-Programming  methods. Second, motivated by the increasing attention paid to measurable robotics and for the sake of reproducibility, data and code (which can be easily adapted to consider any open-chain manipulator) are made available to the community for improvement and to run further comparisons. As indirect but not less important contribution, this paper provides a first physical fully consistent  identified rigid dynamic model of the Jaco robot publicly made available to the robotic community."

Instructions: 

See attached PDF file. Please download and unzip code and data to reproduce our research.