Adaptive Robotic EQ for Well-being (ARoEQ)

Lead Research Organisation: University of Cambridge
Department Name: Computer Laboratory

Abstract

Social robots are envisioned to interact closely with people safely and efficiently, and to add value to people's lives by helping, caring, teaching and entertaining. However, currently there is a major gap between public perception of humanoid / social robot capabilities and their actual capabilities. The cognitive and social capabilities of the current humanoid robots are still very limited.

Although social robotics is an inherently multi-disciplinary field, there are no systematic efforts to develop novel sensing, perception and understanding capabilities for these robots grounded in the state of the art in the fields of affective computing, social signal processing, computer vision and machine learning. To avoid re-inventing the wheel, researchers in HRI often and rightly utilise available sensing / perception tools from other domains, creating their own in-house datasets and evaluations. However, these practices hinder advance in social robotics, leading to a major lack of novel and domain specific tools, and a lack of measures for benchmarking due to a lack of annotated, publicly available multimodal interaction datasets that are vital for comparative evaluation.

This Fellowship aims to address these major gaps in HRI and social robotics. Its vision is to:

(1) equip humanoid robots with novel socio-emotional intelligence and adaptation capabilities grounded in the state of the art in affective computing, social signal processing, computer vision and machine learning fields;

(2) investigate the deployment of humanoid robots as socio-emotionally smart embodied personal devices that can potentially revolutionise our ability to maintain healthier behaviours and working environments, leading to resilient communities.

Planned Impact

The line of research targeted in this funding application is highly interdisciplinary and the research outcomes can be spun into the outer society in various ways. Relevant beneficiaries are identified based on the expected term of impact.

This Fellowship is likely to impact two major areas: Robotics and healthcare. The most immediate application of the proposed research is in the areas of social robotics, telepresence robotics and human-robot interaction. In the short to medium term, the proposed research has the potential to become applicable and expandable to multiple domains including healthcare, social work, education, accommodation, foodservice, retail trade, and arts and entertainment. With advances in AI and physical robot design, robots are starting to take on increasingly complex social roles in homes, workplaces, and public spaces. However, there is a major gap between public perception of humanoid / social robot capabilities and their actual capabilities. The proposed research will improve on the capabilities of the humanoid robots, bringing them closer to the expectations of the public. This is also expected to help with technology adoption, i.e., the adoption and acceptance of social Robotics and Autonomous Systems (RAS) as new products to be used for public good.

Health has been identified as one of the top priorities in the RCUK Strategic Priorities and Spending Plan (2016-2020). In the medium term, the proposed research will contribute to this priority both by developing a better technical understanding of building socio-emotionally intelligent and adaptive robots, and by developing multidisciplinary knowledge about how to deploy such robotic systems as personal and/or self-management technology (e.g., a coach) to strengthen the emotional and social well-being of the unaffected populations for prevention and resilience, both in the workplace and outside.

Academic, artistic and philosophical motivations for researching intelligent machines and robots are well represented by the Science Museum's 2017 exhibition on robots that have started to be used in theatre plays (e.g., Spillikin) and dance performances (e.g., Robot, Blanca Li Dance Company). In the medium to long term, the robotic platform created as part of this project can potentially be used by the performers and artists.

In the longer term, this project will contribute to the UK's competence in Social Robotics and Autonomous Systems (RAS) area adding value into its economy. The project will ensure international visibility for the research team, which in turn will contribute to building the UK's capabilities in social robotics through support for talented people, and attract other investments.
 
Description Main findings from our work published on IEEE Trans. on Cybernetics 2020 [ https://ieeexplore.ieee.org/xpl/tocresult.jsp?isnumber=6352949]
(1) We don't always need sophisticated models as humans rate the machine-like speech to match the machine-like movements better than the more natural movements generated;
(2) Even though the models are not trained with explicit information about subject personalities, subject-dependent learning generates movements that are assessed as more appropriate to the input audio for 'more conscientious' people than for 'less conscientious';
(3) This creates a stepping stone toward learning to synthesize motions for distinctive personality styles rather than manually manipulating a robot's behaviors which is still an open research problem.
Exploitation Route Too early to say as the project started in mid-April 2019.
Sectors Digital/Communication/Information Technologies (including Software),Education,Healthcare

 
Description New Theme on Reproducibility at The 2020 ACM/IEEE International Conference on Human-Robot Interaction
Geographic Reach Multiple continents/international 
Policy Influence Type Influenced training of practitioners or researchers
 
Description Alan Turing Faculty Fellowship
Amount £8,184 (GBP)
Organisation Alan Turing Institute 
Sector Academic/University
Country United Kingdom
Start 04/2019 
End 04/2021
 
Description NHS collaboration 
Organisation Cambridge University Hospitals NHS Foundation Trust
Country United Kingdom 
Sector Public 
PI Contribution Established an in-lab human-robot interaction study setting with a humanoid robot for conducting longitudinal positive psychology therapy sessions. Exploring therapy sessions with a robot is a novel approach for NHS practitioners.
Collaborator Contribution The NHS practitioner provided us guidance in terms of structure of one-to-one sessions, content, measures and questionnaires to be used.
Impact The outcome is the final design of the in-lab human-robot interaction study with a humanoid robot for conducting longitudinal positive psychology therapy sessions.
Start Year 2019
 
Description Festival of Ideas 2019 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Public/other audiences
Results and Impact This was a panel discussion on What Makes Us Human in the Age of AI. The focus was on whether automation make us redundant or are there qualities that are essentially human which will become more sought after. Other panellists included Allegre Hadida, University Senior Lecturer in Strategy at Judge Business School, author and lecturer Laura Dietz and Stephen Cave from the Leverhulme Centre for the Future of Intelligence. Chaired by Julian Clover from Cambridge 105 Radio.
Year(s) Of Engagement Activity 2019
URL https://www.festivalofideas.cam.ac.uk/events/what-makes-us-human-age-ai
 
Description Program Chair for ACM/IEEE International Conference on Human Robot Interaction 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact The 15th Annual ACM/IEEE International Conference on Human Robot Interaction will be held at the Corn Exchange and Cambridge Guild Hall in Cambridge, UK from March 23-26, 2020. I am a Program Chair for this top-tier conference in the area of social robotics and human-robot interaction. The actual impact will be observable after the conference has taken place.
Year(s) Of Engagement Activity 2020
URL https://humanrobotinteraction.org/2020/
 
Description keynote at IEEE FG 2019 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact I gave a keynote talk titled 'Creating Technology with Socio-emotional Intelligence' at IEEE FG 2019, the 14th IEEE Int'l Conference on Automatic Face and Gesture Recognition, held in Lille, France in May 2019.
Year(s) Of Engagement Activity 2019
URL http://fg2019.org/invited-speakers/hatice-gunes/
 
Description seminar at the Centre for Historical Analysis and Conflict Research 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other audiences
Results and Impact I gave an invited seminar titled 'Artificial Emotional Intelligence for Human-Robot Interaction' at the Centre for Historical Analysis and Conflict Research, in the Royal Military Academy Sandhurst with an exgtensive Q&A session regarding the use of AI.
Year(s) Of Engagement Activity 2019