Programming by demonstration of the sequence of tightening a nut allowing variations in tool position

  • José G. Hoyos-Gutiérrez Universidad del Quindío, Armenia
  • Flavio A. Prieto-Ortiz Universidad Nacional de Colombia, Bogotá
Keywords: Robot programming by demonstration, Task segmentation, Complex tasks, Task parameterized Gaussian Mixture model, Petri nets

Abstract

A technique of programming by demonstration of a robot is proposed. Such a technique allows that a robot execute sequential or complex tasks. It uses a combination of Petri nets and task parameterized Gaussian mixture models. The first one handles the task sequence, while the second one allows variations in the position and orientation of objects involved in the task. Using a segmentation task technique, the demonstration is chunked in subtasks. With the subtasks sequence, an action list or plan is obtained and with this, a Petri net is automatically generate. Models of the templates of each subtasks and task parameterized Gaussian mixture models of the subtask that we want to allow variations are also provide to the technique. A function compare one each of the template trajectory with the task parameterized model response trajectory and the most similar indicate that instead of the template, the task parameterized model is use. Through the use of a homemade robot, which executes the task of tightening a nut, the performance of the technique is illustrated by using figures.

Author Biographies

José G. Hoyos-Gutiérrez, Universidad del Quindío, Armenia
MSc. en Ingeniería Eléctrica, Programa de Tecnología en Instrumentación Electrónica, Universidad del Quindío, Armenia
Flavio A. Prieto-Ortiz, Universidad Nacional de Colombia, Bogotá
PhD. en Automática, Facultad de Ingeniería, Universidad Nacional de Colombia, Bogotá
How to Cite
Hoyos-Gutiérrez, J. G., & Prieto-Ortiz, F. A. (2016). Programming by demonstration of the sequence of tightening a nut allowing variations in tool position. TecnoLógicas, 19(36), 77-90. https://doi.org/10.22430/22565337.589

Downloads

Download data is not yet available.
Published
2016-01-30
Section
Research Papers

Most read articles by the same author(s)