Modelo de mezclas gaussianas para tareas de contacto

In this document a probabilistic Gaussian Mixture Model (GMM) to represent contact robotic tasks was developed. We present a comparison between simulation and experimental data for a robot performing simple contact tasks. The simulation includes the manipulator dynamics and a linear model for the contact with the environment. The testing bench consists of (I) a serial manipulator UR3 from Universal Robots, (II) a Force/Torque sensor located at the tip of the manipulator, and (III) a data acquisition system. The GMM was tested using two example tasks. The first task simulates the process of inserting a pen into its cap, and the second task emulates the process of polishing or grinding an object. In both cases, the proposed method was successful according to the AIC and BIC criteria. The results of this project show that this method has a good potential to obtain an implicit model for the proposed contact tasks since it represents the position and force phases during the task and requires few demonstrations.

• Autores: Ascencio Londoño, Sergio Camilo
• Fecha de publicación: 2021
• Institución: Universidad de los Andes
• Repositorio: Séneca: repositorio Uniandes
• Idioma: spa
• OAI Identifier: oai:repositorio.uniandes.edu.co:1992/50931
• Acceso en línea: http://hdl.handle.net/1992/50931
• Palabra clave: Robots móviles; Robótica; Manipuladores (Mecanismo); Sistemas de control inteligente; Sistemas de control de robots; Ingeniería
• Rights: openAccess
• License: https://repositorio.uniandes.edu.co/static/pdf/aceptacion_uso_es.pdf