PolyU Develops Ground-breaking Social Robot to Alleviate Community Mental Health Support Pressure
(中文版在此頁底)
The Hong Kong Polytechnic University officially launched a Theme-based Research Scheme project funded by the University Grants Committee (UGC) to address the increasingly severe mental health issues in Hong Kong. The project, led by Professor Johan F. HOORN from the School of Design and the Department of Computing, has received HK$40.89 million in funding from the UGC and aims to develop a new paradigm for mental health support through the integration of robot technology and artificial intelligence. The UGC's "Theme-based Research Scheme" focuses on leveraging the academic research strengths of UGC-funded universities to investigate themes of strategic importance for Hong Kong's long-term development.
In recent years, the Hong Kong community has faced significant mental health challenges due to social unrest and the pandemic. Research indicates that approximately 12.8% of the population suffers from Post-Traumatic Stress Syndrome, while 61% are experiencing stress, anxiety, and depression. This has not only overwhelmed the mental health care system but has also resulted in billions of dollars in economic losses each year, highlighting the urgent need for innovative solutions.
The project, titled “Social Robots with Embedded Large Language Models Releasing Stress among the Hong Kong Population (T43-518/24-N),” will run for four years until the end of 2029. It aims to utilise localised, cultural, and personal data to train social robots equipped with large language models (LLMs) to provide services to those in need of mental support, particularly vulnerable local communities. These potential users might lack someone to confide in or suitable community support services due to social environment, personal issues, or economic reasons.
The project will employ human-like social robots as natural interaction partners, capable of assisting users with information searches, improving mental and physical health, and supporting educational tasks. The AI system will monitor and utilise topics of concern to citizens to enhance localised and humanised dialogue capabilities. Additionally, it will refine LLMs by strengthening logic, norms, and values to prevent inappropriate behaviour and misinformation. Professor HOORN noted that the team will engage with local communities for field research to assess performance and effectiveness, "we will consult various groups and citizens to inform our robots about cultural considerations and taboos relevant to Cantonese culture, as well as to oversee medical ethics."
These social robots will connect to the award-winning Hong Kong AI hub, HK CityNet, a 2020/21 Theme-based Research Scheme project, while also providing a new software architecture characterised by distributed computing, scalability, and privacy protection. On the hardware side, the team aims to develop affordable, small high-performance computers to run the robots' software at home; the robots will appear as virtual avatars on users' mobile phones, allowing continued communication even when they are out.
The project was announced at an event officiated by Professor Kun-pyo Lee, Dean of the PolyU School of Design, and Professor Qing Li, Head of the PolyU Department of Computing, on 28 February. During the event, Project Leader Professor Hoorn provided an overview of the project to the guests. Co-leaders and collaborators, including Professor Jiannong Cao (PolyU), Professor Sek Ying Chair (CUHK), Professor Song Guo (HKUST), Dr Paul Lo representing Professor Michael Siu (PolyU), Professor Maggie Li (PolyU), Dr Xiaojuan Ma (HKUST), and Professor Chris Webster (HKU) shared their respective approaches to this collaborative initiative. Official collaborators are Professor Bin Xiao (PolyU), Dr Jiayan Pan (Social Work), and Dr Ivy Yan Zhao (PolyU).
香港理工大學正式啟動一項獲大學教育資助委員會（教資會）資助的主題研究計劃，以緩解日益嚴峻的香港心理健康問題。該項目由設計學院及電子計算學系的洪約翰教授（Professor Johan F. HOORN）領導，獲教資會四千零八十九萬港元資助，旨在通過先進技術及人工智能的整合，開展心理健康支援的新格局。教資會的「主題研究計劃」旨在集中本港資助大學的學術研究力量，對香港長遠發展具策略重要性的主題進行研究。
香港社區過去數年經歷過社會事件及疫情，正面臨重大的心理健康挑戰。據研究*顯示，約百分之十二點八的人口患有創傷後壓力症候群，而百分之六十一的人正面對壓力、焦慮和抑鬱，除了導致精神護理系統負荷過重，亦造成每年數十億的經濟損失，因此突顯了創新解決方案的迫切需要。
該名為「香港社區壓力緩解：大型語言模型賦能社交機器人的戰略性研究（T43-518/24-N）」的項目為期四年，直至2029年底，將致力利用本地化、文化和個人數據，對嵌入大型語言模型（LLMs）的社交機械人進行訓練，以向本地有精神支援需要的人士，特別是本地弱勢社群提供服務。此類潛在用戶或受社交環境、個人問題、經濟原因等限制，而缺乏可傾訴的對象或合適的社區支援服務。
項目透過人性化的社交機械人扮演自然的互動夥伴，能夠協助用戶信息搜索、改善身心健康，亦可支援教育工作；其人工智能系統將監測並善用市民關心的話題，以增強本地化和人性化對話能力，同時通過強化邏輯、規範和價值觀來完善大型語言模型，以防止不當行為和虛構「事實」。洪約翰教授指出其團隊將走入本地社區，進行實地研究來評估性能和效果：「我們將諮詢不同的團體及市民，讓他們告訴我們的機械人與廣東文化有關的注意事項和禁忌，亦負責查核有關醫學倫理方面的工作。」
該社交機械人將連接到2020/21年度主題研究計劃項目獲獎的香港人工智能樞紐HK CityNet，同時提供一種新的軟件架構，具備分佈式計算、可擴展性及隱私保護的特點。硬件方面，團隊目標開發價格實惠的小型高性能電腦，用於在家運行機器人的軟件；而該機器人將以虛擬形象出現在用戶手機上，使他們即使外出也能繼續進行溝通。
該項目於2月28日的活動中啟動，由香港理工大學設計學院院長李健杓教授及電子計算學系系主任李青教授主持，而項目負責人洪約翰教授則向與會嘉賓介紹項目概況。出席者包括項目共同負責人包括曹建農教授（香港理工大學）、車錫英教授（香港中文大學）、郭嵩教授（香港科技大學）、盧智恒博士(代表邵健偉教授)（香港理工大學）、李文捷教授（香港理工大學）、麻曉娟博士（香港科技大學）及偉仕達教授（香港大學）。其他參與項目的成員包括肖斌教授（香港理工大學）、趙妍博士（香港理工大學），以及潘佳雁博士（社會工作)。
PolyU Design Research
School of Design
