A Supportive FRIEND at Work; Robotic Workplace Assistance for the Disabled

A Supportive FRIEND at Work; Robotic Workplace Assistance for the Disabled

By Axel Gräser, Torsten Heyer, Leila Fotoohi, Uwe Lange, Henning Kampe, Bashar Enjarini, Stefan Heyer, Christos Fragkopoulos, and Danijela Ristić-Durrant

NOTE: This is an overview of the entire article, which appeared in the December 2013 issue of the IEEE Robotics & Automation Magazine.
Click here to read the entire article.

This article outlines the evolution of an assistive robotic system called the Functional Robot with Dexterous Arm and User-Friendly Interface for Disabled People (FRIEND). The research and development of the assistive robot FRIEND started about 15 years ago. In its first and second generations, different user support scenarios for activities of daily living (ADL) were considered. The first trials on supporting disabled people in a work environment were performed with the third generation, FRIEND III. In the dissemination process of the FRIEND III, it was decided to focus on supporting the integration of disabled people into professional life. The reason is twofold: first, the integration of people with disabilities into professional life increases their self-confidence and life satisfaction; second, the successful implementation will increase interest in such systems from both employers and financial supporters of development. It is nonetheless recognized that implementing FRIEND in a work place increases the complexity of the assistive robotic system because the possible adaptations of the work place environment are limited.

The timeline of the assistive robot FRIEND shows the system’s development over the past decade. (Photos courtesy of IAT, Frank Pusch, and Riad Hamadmad.)

The authors discuss the steps that were necessary to transform FRIEND into a system that supports a quadriplegic end user at work on a daily basis. These steps concern not only the technical adjustments and improvements of the system but also the overcoming of reservations from employers, care personnel, and disabled users. Clarification of details with other organizations such as the department of employment and insurance companies are also critical. The robot usage scenario presented in this article is the support of a quadriplegic individual working as a librarian retrospectively cataloging collections of old books. All the tasks concerning book manipulation such as grasping the book from the book cart and placing it at the specially designed book holder to be read by the end user are carried out autonomously by the system FRIEND, while the cataloging itself is done by the end user using speech recognition. The technical adaptation of the FRIEND system has been driven by the requirement that the robotic system should be very reliable, with little need for unplanned user intervention, and by the requirement that an appropriate standard of safety should be achieved. However, the autonomous functioning of the robot based on vision-based object manipulation cannot always be guaranteed due to a variety of external influences. In the case of failure of the autonomous robot’s task execution, the end user intervenes using an advanced human-machine interface (HMI). A thorough evaluation of FRIEND’s performance in supporting the end user demonstrated that a success rate of 62% was reached for successful fully autonomous task execution, i.e., in 38% of cases, a user intervention was necessary. With the end user intervention, an overall success rate of 95% was achieved. This success rate is an encouraging measure of how FRIEND is able to successfully support the end user in a realistic working scenario.

ABOUT THE AUTHORS

Axel Gräser (ag@iat.uni-bremen.de), Institute of Automation, University of Bremen, Germany.

Torsten Heyer (theyer@iat.uni-bremen.de), Institute of Automation, University of Bremen, Germany.

Leila Fotoohi (fotoohi@iat.uni-bremen.de), University of Bremen, Germany.

Uwe Lange (lange@iat.uni-bremen.de), University of Bremen, Germany.

Henning Kampe (kampe@iat.uni-bremen.de), Institute of Automation, University of Bremen, Germany.

Bashar Enjarini (enjarini@iat.uni-bremen.de), University of Bremen, Germany.

Stefan Heyer (sheyer@iat.uni-bremen.de), Institute of Automation, University of Bremen, Germany.

Christos Fragkopoulos (cfragkopoulos@iat.uni-bremen.de), University of Bremen, Germany.

Danijela Ristić-Durrant (ristic@iat.uni-bremen.de), University of Bremen, Germany.