How can robots and artificial intelligence be used safely to improve the quality of life and increase independent living in an ageing population?
Assistive robots have the potential to provide support for a range of care-related tasks such as physical and social assistance, physiotherapy and rehabilitation. As part of an Innovate UK funded project, CHIRON, the project team have developed a prototype modular overhead/ceiling robotic system that can provide physical and cognitive assistance to frail older adults.
This demonstrator project is investigating and evaluating the safety, regulatory and requirements of close proximate human-robot interaction (HRI) in unstructured domestic environments. It is utilising the CHIRON robotic system within the European Robotics League Certified Anchor Assisted Living Studio at Bristol Robotics Laboratory.
The work is based around a series of experiments designed to validate a range of practical use-cases, derived from potential end-users, occupational and physiotherapists, paid carers and regulators (CQC and MHRA), and potential commercial manufacturing partners. The use-cases will cover functionality and more generalisable HRI aspects such as adaptive and intelligent performance and multi-modal interaction, which will impact on safety relating to human factors, such as: trust, attention, perception, and learning.
A literature review on existing standards and guidelines for assistive robots has been undertaken, in particular with regard to approaches for conducting hazard analysis to evaluate how these approaches relate to the standards and to what extent. The review identified the appropriateness of these for ensuring safety for physically assistive robots.
The project has found that the current assistive robotics standard and hazard analyses do not take into consideration the issues relating to a vulnerable user with accessibility needs (physical and cognitive). They also fail to address issues relating to situations where the user is in direct physical contact with the assistive robot and is using it for physical support while it is moving. The regulations that currently exist in relation to supporting people in the home and in different care settings, including NHS carer guidelines on physically handling patients, have also been reviewed. They highlight that more attention needs to be given to the changing physical and cognitive abilities of the end-user when considering safety and risks.
The team have run workshops with care providers in hospital and residential care settings, with a view to better understanding the issues associated with supporting people with a range of impairments. The outcomes from the workshops have been translated into use cases and scenarios. In combination with Environmental Survey Hazard Analysis (ESHA) the outcomes are being transformed into a set of functional and safety requirements. The next stage is to incorporate the use cases and scenarios into a detailed hazard analysis and design realistic experiments to assess the sensing and safety control requirements of the demonstrator platform.
The CHIRON robot being used in the project is having sensor software and hardware updated. Data is being collected from the installed sensors and this data will be used to develop machine learning algorithms to detect unsafe conditions while a person is using the demonstrator CHIRON system. The unsafe conditions that need to be able to be detected have been identified through the workshops conducted with care providers. The team have been using the data collected to develop the machine learning framework and have developed some preliminary models which can detect the different transitions states during a sit-to-stand activity using the CHIRON system and also detect the transitions patterns of two different people.
The team have also run a number of seminars on their ESHA studies, both on the methodology and its application, and have been developing practical guidance to be published in the Body of Knowledge.
A completed definition of a systems architecture model-based Distributed Adaptive Control (DAC) has also been completed. The DAC is a standard architecture reference model for autonomous systems which provides a template for identifying the functions of the robot to which machine learning can be applied.
Presentations and papers
- March 2020 - a tutorial on Environmental Survey Hazard Analysis (ESHA) was presented at the European Robotics Forum in Malaga, Spain.
- February 2020 - poster presentation at Safety-Critical Systems Symposium. Using Task Analysis and Environmental Survey Hazard Analysis to identify
requirements of autonomous systems (poster abstract), Chris Harper, Daniel Delgado Bellamy, Sanja Dogramadzi S., Praminda Caleb-Solly.
- February 2020 - poster presentation at Safety-Critical Systems Symposium. Safety Aspects of Complex Human-Robot Interaction in Healthcare Robotics
(poster abstract), Daniel Delgado Bellamy, Chris Harper, Sanja Dogramadzi S.
Praminda Caleb-Solly, Proc. 28th Safety-critical Systems Symposium, York UK,
February 2020, ISBN 9781713305668, pp417-420
- 12 September 2019 - Praminda Caleb-Solly presented a session at the National Association for Safety and Health in Care Services annual conference exploring safety-related issues.
- 12 July 2019 - Praminda Caleb-Solly co-organised and presented a session on ensuring user acceptance, usability and safety of physically assistive robots through user-centred design, at the British Society of Gerontology in Liverpool.