Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Fast Contact Detection and Classification for Kinesthetic Teaching in Robots using only Embedded Sensors

Salt Ducaju, Julian LU orcid ; Olofsson, Björn LU ; Robertsson, Anders LU and Johansson, Rolf LU orcid (2022) 31st IEEE International Conference on Robot and Human Interactive Communication (RO-MAN) p.1138-1145
Abstract
Collaborative robots have been designed to perform tasks where human cooperation may occur. Additionally, undesired collisions can happen in the robot’s environment. A contact classifier may be needed if robot trajectory recalculation is to be activated depending on the source of robot–environment contact. For this reason, we have evaluated a fast contact detection and classification method and we propose necessary modifications and extensions so that it is able to detect a contact in any direction and distinguish if it has been caused by voluntary human cooperation or by accidental collision with a static obstacle for kinesthetic teaching applications. Robot compliance control is used for trajectory following as an active strategy to... (More)
Collaborative robots have been designed to perform tasks where human cooperation may occur. Additionally, undesired collisions can happen in the robot’s environment. A contact classifier may be needed if robot trajectory recalculation is to be activated depending on the source of robot–environment contact. For this reason, we have evaluated a fast contact detection and classification method and we propose necessary modifications and extensions so that it is able to detect a contact in any direction and distinguish if it has been caused by voluntary human cooperation or by accidental collision with a static obstacle for kinesthetic teaching applications. Robot compliance control is used for trajectory following as an active strategy to ensure safety of the robot and its environment. Only sensor data that are conventionally available in commercial collaborative robots, such as joint-torque sensors and joint-position encoders/resolvers, are used in our method. Moreover, fast contact detection is ensured by using the frequency content of the estimated external forces, whereas external force direction and sense relative to the robot’s motion is used to classify its source. Our method has been experimentally proven to be successful in a collaborative assembly task for a number of different experimentally recorded trajectories and with the intervention of different operators. (Less)
Please use this url to cite or link to this publication:
author
; ; and
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
keywords
Robotics, Contact Detection, Kinesthetic Teaching
host publication
Proc. 2022 31st IEEE International Conference on Robot and Human Interactive Communication (RO-MAN) Aug 29 - Sep 2, 2022
pages
8 pages
conference name
31st IEEE International Conference on Robot and Human Interactive Communication (RO-MAN)
conference location
Naples, Italy
conference dates
2022-08-29 - 2022-09-02
external identifiers
  • scopus:85140731049
ISBN
978-172818859-1
DOI
10.1109/RO-MAN53752.2022.9900800
project
Human-Robot Collaboration for Kinesthetic Teaching
RobotLab LTH
ELLIIT LU P06: Collaborative Robotic Systems
WASP: Wallenberg AI, Autonomous Systems and Software Program at Lund University
language
English
LU publication?
yes
id
8ef803eb-edf6-4e6d-b899-400ba9fbafe4
date added to LUP
2022-09-02 20:00:12
date last changed
2024-03-08 14:32:40
@inproceedings{8ef803eb-edf6-4e6d-b899-400ba9fbafe4,
  abstract     = {{Collaborative robots have been designed to perform tasks where human cooperation may occur. Additionally, undesired collisions can happen in the robot’s environment. A contact classifier may be needed if robot trajectory recalculation is to be activated depending on the source of robot–environment contact. For this reason, we have evaluated a fast contact detection and classification method and we propose necessary modifications and extensions so that it is able to detect a contact in any direction and distinguish if it has been caused by voluntary human cooperation or by accidental collision with a static obstacle for kinesthetic teaching applications. Robot compliance control is used for trajectory following as an active strategy to ensure safety of the robot and its environment. Only sensor data that are conventionally available in commercial collaborative robots, such as joint-torque sensors and joint-position encoders/resolvers, are used in our method. Moreover, fast contact detection is ensured by using the frequency content of the estimated external forces, whereas external force direction and sense relative to the robot’s motion is used to classify its source. Our method has been experimentally proven to be successful in a collaborative assembly task for a number of different experimentally recorded trajectories and with the intervention of different operators.}},
  author       = {{Salt Ducaju, Julian and Olofsson, Björn and Robertsson, Anders and Johansson, Rolf}},
  booktitle    = {{Proc. 2022 31st IEEE International Conference on Robot and Human Interactive Communication (RO-MAN) Aug 29 - Sep 2, 2022}},
  isbn         = {{978-172818859-1}},
  keywords     = {{Robotics; Contact Detection; Kinesthetic Teaching}},
  language     = {{eng}},
  pages        = {{1138--1145}},
  title        = {{Fast Contact Detection and Classification for Kinesthetic Teaching in Robots using only Embedded Sensors}},
  url          = {{https://lup.lub.lu.se/search/files/160718393/ROMAN2022_reviewed_2_.pdf}},
  doi          = {{10.1109/RO-MAN53752.2022.9900800}},
  year         = {{2022}},
}