澳大新語•2021 UMAGAZINE 24 1
Publisher: University of Macau Chief Editor: Katrina Cheong Deputy Chief Editor: Ella Cheong Editors: Davis Ip, Debby Seng Translators: Ruby Chen, Anthony Sou Advisors: Timothy Simpson, Associate Professor, Department of Communication, Faculty of Social Sciences Tang Keng Pan, Professor Emeritus, Department of Chinese Language and Literature, Faculty of Arts and Humanities Lampo Leong, Distinguished Professor and Director, Centre for Arts and Design, Faculty of Social Sciences Design: Jack Ho Address: Room G012, Administration Building, University of Macau, N6, Avenida da Universidade, Taipa, Macau, China Contact: Tel: (853) 8822 8833 Fax: (853) 8822 8822 Email: prs.publication@um.edu.mo Printing: Hamah (Macau), Limitada ISSN: 2077‑2491 Published biannually since 2009, UMagazine aims to report great ideas and research breakthroughs from the University of Macau. It also showcases the latest developments and achievements of the university in teaching, research, and service. 出版: 澳門大學 總編輯: 張惠琴 副總編輯: 張愛華 編輯: 葉浩男、盛惠怡 翻譯: 陳靜、蘇恩霆 顧問: 社會科學學院傳播系副教授Timothy Simpson 人文學院中國語言文學系榮休教授鄧景濱 社會科學學院藝術設計中心主任/特聘教授梁藍波 排版: 何杰平 地址: 中國澳門氹仔大學大馬路澳門大學N6行政樓G012 室 聯絡: 電話: (853) 8822 8833 傳真: (853) 8822 8822 電郵: prs.publication@um.edu.mo 製版印刷: 澳門豪邁實業有限公司 國際刊號: 2077‑2491 《澳大新語》創於2009年，為澳門大學官方刊物， 每年出版兩期，旨在展示澳門大學的創見和突破、 報導教研和社會服務的最新發展和成果。 2021年 | 總第24期
過去40年，一代代澳大人不懈研發創 新科技，並致力將它們轉化為改善人 們生活的產品和服務。澳大目前在中 醫藥質量、芯片設計、智慧城市物聯 網、精準醫學、先進材料、大數據、人 工智能等方面已有可喜成績。 當前澳門經濟急需轉型，科技發展更 見重要，澳大作為澳門唯一的綜合性 公立大學，近年積極落實習近平主席 有關「創造更多科研成果」的指示，緊 扣國家發展戰略、配合澳門特區政府 施政部署，全力推動科技成果轉化， 促進澳門產業多元化發展。 約30年前，澳大已開展人工智能方 面的教學和研究，近年更與海內外 院校和企業加強合作，攜手研發真 正惠及所有人的技術、產品和服務。 今期《澳大新語》聚焦大學多個人工 智能科研項目及其應用，包括自動 巴士、無人船、工業機器人、智慧旅 遊和偽造圖像偵測，全部切合澳門 發展所需，也展現了澳大為推進粵 港澳大灣區建設國際科技創新中心 的不懈努力。 我們也介紹了澳大另外兩個重要科 研方向，分別是制訂在國際性藥典 的中藥質量標準和區域海洋研究。 此外，我們也專訪了社會科學學院 院長胡偉星教授、哲學與宗教學教 授王慶節。在「學術研究」專欄，澳大 學者介紹新冠疫情與犯罪，以及近 年推動哥德文學研究的進展。 Over the past four decades, generations of researchers at the University of Macau (UM) have developed many innovative technologies and transformed them into products and services that can improve people’s lives, most notably in such fields as Chinese medical sciences, internet of things for smart cities, chip design, precision medicine, advanced materials, big data, and artificial intelligence. Indeed, not only has technological innovation become ever more important to Macao in view of the pressing need to upgrade the city’s industrial structure, but it has also been elevated to the status of national strategy, as is reflected in President Xi Jinping's explicit instructions for higher education institutions in Macao to ‘attain more achievements in scientific research’. As the only public comprehensive university in the city, UM actively carries out President Xi’s instructions and closely aligns itself with national development strategies and the Macao SAR government’s policies, in order to promote technology transfer and the diversification of Macao’s economy. This issue of UMagazine focuses on the AI projects conducted by professors at UM, which began its AI education and research three decades ago. In recent years, we have strengthened our partnerships with universities and companies at home and abroad to develop new AI technologies, products, and services that will truly benefit everyone. We spotlight some of our most exciting projects, including a self-driving bus, unmanned marine vessels, industrial robots, and software for smart tourism and image forgery detection. These projects, which will find broad applications in Macao, prove our effort to support the GuangdongHong Kong-Macao Greater Bay Area’s transformation into a global technology and innovation hub. In this issue, we also bring you stories about two important research areas: establishing quality standards for Chinese medicinal herbs for major pharmacopoeias and regional oceanology. Other stories not to be missed include interviews with Faculty of Social Sciences Dean Prof Richard Hu and Prof Wang Qingjie in the Department of Philosophy and Religious Studies, as well as two articles in the ‘Academic Research’ column, which discuss the impact of the COVID-19 on crime and its control, and the latest developments in Gothic scholarship at UM. 編者的話 張惠琴 Katrina Cheong EDITOR’S WORDS
CONTENTS 目錄 2021年 | 總第24期 Autumn / Winter 2021 | Issue 24 人工智能新技術研發自動駕駛 智能海洋機器人用途廣泛 智能機器人創新惠民 智能偵測圖像篡改 社交媒體數據驅動智慧旅遊 制訂標準助推中醫藥國際拓展和澳門產業發展 Unleashing the Power of AI on Self-driving Smart Marine Robots to See Greater Use Intelligent Robots Serve the Society with Innovations The AI Image Forgery Detective Social Data Drives Smart Tourism Establishing Quality Standards to Promote Internationalisation of Chinese Medicine and Chinese Medicine Industry in Macao 06 12 16 20 24 28 封面專題 COVER STORY 專題探討 TOPIC INSIGHT
區域海洋研究監測自然災害 新冠疫情與犯罪及其治理 在澳門推動哥德文學研究 胡偉星：培養數據時代的公共行政人才 王慶節談哲學與人生 34 40 46 52 55 Regional Oceanology Monitors Natural Disasters The COVID-19 Pandemic and Its Effect on Crime and Its Control Developing Gothic Scholarship in Macao Richard Hu - Nurturing Public Administration Professionals in the Big Data Era Wang Qingjie on Philosophy and Life 學術研究 ACADEMIC RESEARCH 人物專訪 EXCLUSIVE INTERVIEW
COVER STORY • 封面專題 2021 UMAGAZINE 24 • 澳大新語 6 人工智能新技術研發自動駕駛 Unleashing the Power of AI on Self‑driving 文/葉浩男‧圖/何杰平、部分由受訪者提供 Chinese & English / Davis Ip ‧ Photo / Jack Ho, with some provided by the interviewee 澳門第一台自動駕駛巴士於2020年底啟用 Macao’s first self‑driving bus was launched at UM in late 2020 科幻電影中自動車穿梭大都會街頭的畫面，再也不 是海市蜃樓。為了研究自動駕駛技術，澳門大學正與 本地和內地多個機構開展一個大型科研項目，並在 澳大校園啟用澳門第一台自動駕駛巴士，作為測試 新技術的平台。 以自動駕駛巴士測試新技術 這架自動駕駛巴士在2020年10月啟用，有八個座位 和六個站位，時速最高40公里，是科研項目「協同智能驅 動的無人駕駛關鍵技術與平台研究」的重要一環。該項 目2019年起獲澳門科學技術發展基金資助，由智慧城市
澳大新語 • 2021 UMAGAZINE 24 7 封面專題 • COVER STORY 物聯網國家重點實驗室（澳門大學）、中國科學院深圳先 進技術研究院、國防科技大學、百度和深圳海梁科技共 同承擔，並獲澳門電訊提供流動網絡技術支援。 研究團隊由澳大科技學院院長、電腦及資訊科技系 講座教授須成忠領導。須教授研究自動駕駛已有十 多年，在美國底特律任教時曾與通用汽車等大企業 合作研究智能駕駛，2011年回國後繼續探索相關技 術，2019年加入澳大後不久開展這個自動駕駛項目。 全球各地都在研究自動駕駛。今日的自動車已能在受 控制的環境下正常行駛，但現有技術還不足以製造出 能安全、恰當地應對極端天氣和其他突發情況的全自 動汽車，因此須教授的團隊正在研究令自動駕駛更安 全的技術，成果令人鼓舞。單在2019年，他們就有22 篇論文獲國際學術期刊登載，也有在一些頂尖國際會 議上發表論文。 提升自動車決策能力 卷積神經網絡是一種主流的機器學習模型，某程度模 仿人腦的感知方式，是自動駕駛技術的重要一環。須 教授說，我們可以用大量關於交通情況的圖像數據來 訓練一個卷積神經網絡，使它能在新的圖像中辨別出 物件、車輛和行人。 卷積神經網絡的決策能力很取決於它用來學習的訓 練數據。目前多數用於訓練自動車的數據都是在良好 天氣下收集所得，因此自動車平日能比較容易檢測出 物件，但遇上颱風等惡劣天氣和其他特殊情況時卻很 可能失靈。我們原則上可以引入更多數據，為每種路
COVER STORY • 封面專題 2021 UMAGAZINE 24 • 澳大新語 8 況都來一個訓練卷積神經網絡（或其他機器學習模型）， 但實踐中卻難以有足夠的數據和時間來訓練這些網 絡，也無法預見汽車所有可能遇到的情況。 應對這個問題的方法之一就是運用「遷移學習」人工智 能方法，原理有點像訓練一個單車手學習駕駛電單車。 單車手憑著他們的經驗，可能駕駛電單車時會比其他 人更易保持平衡，毋須像其他人一樣從頭學起。同樣， 如果沒有足夠的數據或時間，我們可以用已經訓練好 的機器學習模型為基礎，為類似任務訓練出新的模型。 為了提升遷移學習的成效，澳大與百度的研究 人員提出了一種新算法，名為Re‑Initializing the Fully‑connected LayEr（簡稱RIFLE），用於訓練基於 遷移學習方法的卷積神經網絡時的「反向傳播」過程。 他們先用這個新算法訓練一個卷積神經網絡，然後以 它來分類、檢測和分割數以萬計的圖像，發現其表現 遠勝一些經主流算法訓練的對手。相關論文已在頂尖 的國際機器學習年會上發表。 加快訓練機器學習模型 重新訓練一個卷積神經網絡前，我們可以先刪除一些 與新任務無關的部分，這個過程稱為「剪枝」。「剪枝」 有助降低卷積神經網絡的複雜程度、從而加快運算， 但有時也會削弱其執行任務的能力。 針對這個問題，澳大、中科院深圳先進技術研究院和 百度的研究人員開發了一種新的「剪枝」方法，名為 Attentive Feature Distillation and Selection（簡稱 AFDS）。他們用AFDS和多種主流算法修剪一個有101 個卷積層的卷積神經網絡，再用這些網絡來分析六個 圖片數據庫，辨識圖片上有甚麼物件。他們發現，在運 算量下降30%的情況下，用AFDS修剪的網絡達到幾 乎同樣準確的辨識結果；即使運算量下降90%，準確 度仍能夠保持在約70%，遠高於用其他方法修剪的網 絡。相關論文已在頂尖的國際學習表徵年會上發表。 自動車通常會用卷積神經網絡來分析周邊的物體，它 們一旦受到網絡攻擊或遇到不太清晰的物件（例如有 塗鴉的路標）時很易判斷錯誤，例如將停車標誌誤判 為讓先標誌，隨時釀成慘劇。因此，研究人員開發了 LAFEAT算法，令卷積神經網絡面對攻擊或噪聲時更 具魯棒性（robustness，又譯穩健性）。這款新算法在 實驗中的表現遠勝10多種現有的算法。 在2021年中，澳大和百度的研究人員將LAFEAT算法 澳大自動駕駛研究團隊 The UM research team behind the self‑driving project
澳大新語 • 2021 UMAGAZINE 24 9 封面專題 • COVER STORY a) Pre-trained model … pool classifier f (N ) f (1) … pool xN xN−1 x1 b)Train additional logits f (N ) f (1) … pool classifier x xN xN−1 … pool L(·,y) L1 pool L(·,y) LN−1 ( ( f (N ) f (1) x−xˆ arg max yˆ = arg maxy c) Generate adversarial examples with additional logits classifier … pool pool + L(·,y) Ladv − … ( o if p p λ1 x y d)Attack pre-trained model with adversarial examples … pool classifier yˆ xˆ f (1) f (N ) x1 g(1) g(N−1) g(1) g(N−1) xˆ AFDS算法有助提升機器學習模型的運算速度 The AFDS algorithm accelerates the training of machine learning models 帶來額外挑戰，所以他的團隊亦在研究應對澳門複 雜路況的技術。 那麼，到底我們何時才能在日常生活用自動車出行？ 須教授說，要真正廣泛使用自動車，我們還要克服不少 技術難題，也需要有新的道路、網絡基建和法律法規配 合，有賴社會各界共同努力。「我們會不斷開發新的自 動駕駛技術。通過在澳大啟用自動巴士，我們也希望提 升公眾對自動車的認識，為在澳門實現自動駕駛創造 有利條件。」 在國際計算機視覺與模式識別會議上發表，該會議的 論文錄取率僅為4.59%。他們也用這項算法參加了美 國伊利諾大學、清華大學和阿里安全合辦的「CVPR安 全AI挑戰者賽」算法比賽，與全球1680支隊伍切磋，勇 奪亞軍，成為唯一來自澳門的獲獎隊伍。 走向自動駕駛的未來 研究團隊亦在深入研究人機交互技術，重點包括自 然語言處理，目標是讓自動車正確回應語音指令。須 教授也說，澳門路面較窄、電單車特別多，對自動車 須成忠教授 Prof Xu Chengzhong Once only visible in science fiction films, self‑driving cars roaming streets in futuristic cities may enter your car park sooner than you think. To advance the technology behind such cars, the University of Macau (UM) and its partners from Macao and mainland China are conducting a major research project. Together, they have launched Macao’s first‑ever self‑driving bus as a testing vehicle at the university. Tests Underway on an Unmanned Bus The self‑driving bus hit the road in October 2020. It has eight seats, allows six standees, and can travel up to 40 km per hour. Its launch is part of the 'Key Technologies and Platforms for Collaborative Intelligence‐driven Autonomous Vehicles’ project, which has been supported by the Macao Science and Technology Development Fund since 2019. The researchers come from the State Key Laboratory of Internet of Things for Smart City (University of Macau), the Shenzhen Institutes of Advanced Technology (SIAT) under the Chinese Academy of Sciences, the National University of Defense Technology, Baidu, and Shenzhen Haylion Technologies. Macao’s telecommunications service provider CTM provides mobile network technology to support the project. The research team is headed by Prof Xu Chengzhong, dean of UM’s Faculty of Science and Technology and chair professor of computer and information science. Prof Xu has studied self‑driving technology for over a decade. He worked with American automakers such as General Motors on the development of intelligent vehicles when he was a professor in Detroit. He has continued to explore the field since returning to China in 2011, and kick‑started this self‑driving car project in Macao after he joined UM in 2019.
COVER STORY • 封面專題 2021 UMAGAZINE 24 • 澳大新語 10 Self‑driving research has proliferated across the world. Today, cars can already drive themselves smoothly in controlled environments. However, Prof Xu says that we are still far from having fully autonomous cars that can safely navigate in extreme weather conditions and other unpredictable situations in the real world. In making self‑driving cars safer, Prof Xu’s team has seen some encouraging results. In 2019 alone, they published 22 papers on related technologies in leading academic journals and presented some other papers at top international conferences. Teaching Cars to Make Smart Decisions According to Prof Xu, Convolutional Neural Network (CNN), which is a popular type of machine learning model, is essential to self‑driving cars. A CNN loosely mimics the way a human thinks. Given a large image dataset for it to learn from, a CNN can be trained to detect objects such as road signs, vehicles, and pedestrians from new images. The decision‑making power of a CNN largely depends on its training data. Most existing data for training self‑driving cars was collected in clear weather conditions, so it is easier for self‑driving cars to detect objects in good weather, but they may make serious mistakes in adverse weather conditions such as typhoons and snow. In theory, scientists can bring in more data to train a CNN (or other machine learning model) for every scenario. In practice, however, they probably won’t have enough data or time, nor can they foresee all the situations a car might encounter. One of the solutions is transfer learning, which works a bit like a cyclist learning to ride a motorcycle. By drawing on their experience, cyclists may find it easier to balance themselves on two wheels, without starting from scratch. Similarly, in the absence of enough data or time, researchers can train a machine learning model based on a ‘transferred’ model that has already been perfected for a similar task. With the self‑driving car project, researchers at UM and Baidu have proposed a new way to help CNNs based on transfer learning to accurately detect objects like stop signs. Their algorithm is called Re‑Initialising the Fully‑connected LayEr (RIFLE), which is used in the ‘back propagation’ process in the training of CNNs. The researchers used a CNN trained with the new algorithm to classify, detect, and segment tens of thousands of images. It turns out that this CNN outperforms its rivals trained with some mainstream algorithms. The team has presented a paper about 研究人員可在設於智慧城市物聯網國家重點實驗室（澳門大學）的實驗平台用模型車輛測試自動駕駛技術 This testbed in the State Key Laboratory of Internet of Things for Smart City (University of Macau) allows researchers to test their self‑driving strategies on mini‑cars in a mock urban scenario
澳大新語 • 2021 UMAGAZINE 24 11 封面專題 • COVER STORY the RIFLE algorithm at the prestigious International Conference on Machine Learning. Faster Training of Learning Models We can remove unnecessary parts of a CNN before retraining it for a new but similar task. Known as ‘pruning', the process can reduce the size of the original network and thus shorten the time required for training the new one. However, pruning sometimes sharply weakens the network’s decision‑making power. To address this problem, researchers at UM, SIAT, and Baidu proposed a new pruning method called ‘Attentive Feature Distillation and Selection’ (AFDS). They used ADFS and some mainstream counterparts to prune a 101‑layer CNN, and then used the final products to identify objects from images stored in six databases. The results show that the AFDS‑pruned network remains very effective despite a 30 per cent cut on the amount of computation. Even when the amount of computation was reduced by 90 per cent, the network accuracy rate stood at 70 per cent, much higher than its competitors. The researchers have presented an academic paper about the AFDS method at the influential International Conference on Learning Representations. Self‑driving cars must be able to identify everything around them. CNNs can usually perform such tasks effectively, but they can be vulnerable when facing cyberattacks or unclear objects like road signs with graffiti. This can cause potentially fatal mistakes like misreading a stop sign as a give‑way sign. Therefore, Prof Xu’s team has developed the ‘LAFEAT’ algorithm, which can make CNNs more robust to adversarial attacks or noise. The new algorithm has outperformed a dozen existing options in computer trials. In mid‑2021, researchers at UM and SIAT presented the LAFEAT algorithm at the Conference on Computer Vision and Pattern Recognition, which has an acceptance rate of just 4.59 per cent. They also used the new algorithm to compete with 1680 teams from around the world at the CVPR Security AI Challenger, an algorithm competition jointly organised by the University of Illinois, Tsinghua University, and Alibaba Security. The team won a second prize, becoming the only winning team from Macao. Stepping into a Driverless Future Prof Xu’s team is also studying technologies to enable better interactions between humans and self‑driving cars. For instance, they are designing natural language processing solutions that will allow cars to respond appropriately to spoken commands. Moreover, they are searching for ways to help cars adapt to the complicated road conditions in Macao, a city known for its narrow roads and abundance of motorbikes. So, when can we take our hands off the wheels? Prof Xu believes that many technological breakthroughs need to be achieved before self‑driving cars can finally go mainstream. A driverless future would also need major transformations in transport and telecommunications infrastructure as well as new legal rules ‑ all requiring the concerted effort of various sectors of the community. Prof Xu says, ‘We will of course continue to advance self‑driving technology, but by launching our bus on the UM campus, we also hope to raise the public’s understanding of self‑driving cars, so that we can be better prepared for their deployment in Macao.’ 卷積神經網絡受攻擊時或會誤判路標 Convolutional neural networks under cyberattacks may misjudge road signs 掃二維碼 觀看訪談片段 Scan the QR code to watch the interview
COVER STORY • 封面專題 2021 UMAGAZINE 24 • 澳大新語 12 文/葉浩男‧圖/何杰平、部分由受訪者提供 Chinese & English / Davis Ip ‧ Photo / Jack Ho, with some provided by the interviewee Smart Marine Robots to See Greater Use 智能海洋機器人用途廣泛 Carlos Silvestre教授（中）、研究助理Joel Reis博士（左）和博士生余甘 Prof Carlos Silvestre (middle), research assistant Dr Joel Reis (left), and PhD student Yu Gan 澳門大學電機及電腦工程系Carlos Silvestre 教授的團隊正在研發新型自主船艇。他們在 2020年一項國際智能無人船賽事奪冠，目前則 在研究智能水下無人艇，能用於執行海底測繪 等諸多任務。 應用廣泛 Silvestre教授在2011年領導成立「基於傳感器的協作 機器人研究實驗室」（SCORE實驗室），目前與研究助 理Joel Reis博士、博士生余甘和其他學生一同開展動 力系統理論的科學探索，並且運用研究成果開發智能 海洋機器人和空中無人機。
澳大新語 • 2021 UMAGAZINE 24 13 封面專題 • COVER STORY 他們在比賽期間編寫了智能製導、控制和導航程 序並安裝在USV上，然後在海上執行自動導航、避 開障礙物和目標識別等任務。他們的USV最終的 追蹤準確度不僅遠高於十名對手，更被賽事主辦 方譽為「史無前例」。 新型無人潛水器 此外，Si l vest re教授的團隊正在設計一種新型自 主水下航行器，將會配備專業級的聲納系統、攝像 頭和慣性傳感器。這個項目名為「ORVIS‑Ocean Robotic Vehicles for Intervention in Shallow Waters」，為期數年，2020年起由澳門科學技術發 展基金資助。 據Si lvest re教授介紹，這款新型水下機器人將會 配備聲納傳感器，在低能見度的水底仍能產生高 解像度的圖像。「在颱風等天災後，我們可以使用 這個機器人來檢查澳門能見度很低的水域的水下 設施有否損壞。海洋機器人可以巡邏水域、檢查海 堤等海洋設施、繪製海床地圖和監測海洋生態系 統，相信將會在澳門得到更廣泛的應用。」 Silvestre教授早在30多年前開始研究海洋機器人。當 時他是里斯本大學高等技術學院的碩士生。他說，在 不同類型的機器人中，無人水面艦艇（USV）的用途尤 其廣泛，在執行巡邏、貨運、海洋研究、搜救、油氣勘探 和海底電纜安裝等任務時均有明顯優勢。 設計導航及控制算法 由於風、浪、洋流等環境因素，要使USV始終在正 確時間沿著正確路徑航行並非易事。USV需要一 個由複雜算法組成的軌跡跟蹤控制系統，才能即 時、準確和安全地運作，並在偏離預定路徑時進行 快速修正。 SCORE實驗室近年來在設計非線性軌跡跟蹤智 能控制系統方面取得豐碩成果，它們的有效性已 在電腦仿真和澳大校園內的實地試驗得到驗證。 實驗室成員開發的算法能讓USV避開障礙物、在 電力耗盡時返回出發點和應對其他突發情況。 SCORE實驗室成員憑藉他們的算法，在2020年 底首屆珠海萬山國際智能船艇公開賽獲得冠軍。 SCORE實驗室在澳大校園的湖面試測無人船 The SCORE Laboratory’s unmanned vessel is being tested on the UM campus 澳大團隊在首屆珠海萬山國際智能船艇公開賽的 一個項目奪冠 The UM team wins a first prize at the first Zhuhai Wanshan International Intelligent Vessel Competition
COVER STORY • 封面專題 2021 UMAGAZINE 24 • 澳大新語 14 Diverse Applications In 2011, Prof Silvestre founded the Sensor‑based Cooperative Robotics Research Laboratory (SCORE Laboratory), where he conducts scientific investigations in dynamical systems theory with research assistant Dr Joel Reis, Yu Gan, and several other PhD students. Moreover, they apply their research results to the development of advanced intelligent marine robots and aerial drones. A team led by Carlos Silvestre, a professor in the University of Macau’s (UM) Department of Electrical and Computer Engineering, is developing next‑generation marine vessels that can perform tasks on their own. Since winning a first prize at an international intelligent vessel competition, the team has been designing underwater robots which can, among many other things, map the ocean floor. 無人船裝有一個軌跡跟蹤控制系統 The unmanned vessel is equipped with a trajectory tracking control system
澳大新語 • 2021 UMAGAZINE 24 15 封面專題 • COVER STORY Innovative Unmanned Underwater Vessels Moreover, Prof Silvestre’s team is designing an innovative autonomous underwater vehicle, which will be equipped with high‑end, professional‑level sonar systems, cameras, and inertial sensors. This multi‑year project, titled ‘ORVIS ‑ Ocean Robotic Vehicles for Intervention in Shallow Waters’, has been funded by the Macao Science and Technology Development Fund since 2020. According to Prof Silvestre, the new underwater robot will have sonar sensors that can produce high‑resolution underwater images even in low‑visibility conditions. ‘After typhoons or other natural disasters, it will be possible to use the vessel to check whether underwater facilities are damaged in the waters of Macao where visibility is quite low. I believe Macao will see broader applications of marine robots, which can patrol waters, inspect marine facilities like seawalls, map the seabed, and monitor marine ecosystems.’ Prof Silvestre began studying marine robotics more than three decades ago, when he was still a master’s student in Instituto Superior Técnico at the University of Lisbon. He says that, among different types of robots, unmanned surface vessels (USVs) are particularly useful in activities such as patrolling, cargo shipping, ocean research, maritime search and rescue, oil and gas surveying, and installation of submarine cables. Designing Navigation and Control Algorithms Due to wind, waves, ocean currents, and other environmental factors, it is not easy for USVs to always move along the right path at the right time. It takes a trajectory tracking control system, consisting of sophisticated intelligent algorithms, to accurately and safely maneuver a USV in real time, and to make quick corrections when the vehicle deviates from the intended path. In recent years, the SCORE Laboratory has made considerable progress in designing non‑linear trajectory tracking intelligent control systems, whose effectiveness has been proved in computer simulations and actual trials with a USV on the UM campus. Laboratory members have also developed algorithms that allow USVs to avoid obstacles, return to their start locations when running out of power, and cope with unexpected situations. With their cutting‑edge algorithms, the SCORE laboratory members won a first prize at the first Zhuhai Wanshan International Intelligent Vessel Competition in late 2020. During the competition, the UM team wrote an intelligent guidance, control and navigation programme and installed it on a USV, which then performed various tasks on the sea, such as automatic navigation, obstacle avoidance, and target identification. The tracking accuracy achieved by the UM team was not only far greater than the results obtained by all ten of its rivals in the competition, but was also praised as ‘unprecedented’ by the event organisers. Carlos Silvestre 教授 Prof Carlos Silvestre 掃二維碼 觀看訪談片段 Scan the QR code to watch the interview
COVER STORY • 封面專題 2021 UMAGAZINE 24 • 澳大新語 16 文/葉浩男‧圖/何杰平、部分由受訪者提供 Chinese & English / Davis Ip ‧ Photo / Jack Ho, with some provided by the interviewee 智能機器人創新惠民 Intelligent Robots Serve Society with Innovations 消毒機器人協助防疫 2020年新冠疫情初期，徐教授的團隊獲澳門科學技 術發展基金資助開發了智能機器人「消毒智多星」， 期間獲澳門發展及質量研究所提供空氣指標監測 支援。他說消毒機器人能減省人力、提高效率、降低 醫護人員接觸病毒的風險和確保消毒劑均勻噴灑。 隨著人工智能的進步和行業需求的增長，智能 機器人早已成為全球各地的研究熱點。澳門大 學機電工程系教授徐青松一直帶領團隊研發創 新的智能機器人，服務民生和社會。 徐青松教授（中）的團隊研製各類機器人 The team of Prof Xu Qingsong (middle) has designed many types of robots
澳大新語 • 2021 UMAGAZINE 24 17 封面專題 • COVER STORY 智能工業機器臂 An intelligent industrial robotic arm 微操作機器人促進生物醫學工程 徐教授團隊早前還研發了智能微操作機器人系統。 該系統能夠自動在微納米尺度上操控微注射器和 微夾鉗，可以用於基因編輯等活體細胞操作，大大 增加細胞在顯微注射後的存活率，使顯微注射更可 靠、效果更穩定，滿足生物醫學工程對活體細胞操 作不斷增長的需求。他們還發明了新型智能精確運 動與力度混合控制算法，讓微操作機器人工作時更 迅速和更準確。 此外，徐教授團隊的項目「機器人微夾鉗系統研發 及産業應用」是澳門大學—華發集團聯合實驗室的 首批科創項目之一，正在珠海澳大科技研究院的支 援下，將研究成果轉化為面向市場的產品。 經過10餘年研發積累，徐教授先後擔任多份著名國 際期刊的編委，並多次獲得澳門科學技術獎勵，期 望團隊再接再厲：「我們也正在開發血管機器 人、高空作業機器人等智能機器人，進一步以科 技創新改善人們的生活。我相信智能機器人將 會與人共融，更好地服務社會。」 徐教授說：「我們一手研發了機器人的零件和程式， 更包辦其外觀設計和組裝，所以能夠壓縮成本和售 價，吸引更多機構使用機器人。」他表示，澳大已就 相關技術申請專利，也正與企業洽談專利權轉讓， 期望將智能消毒機械人批量生產，推向粵港澳大灣 區等地的市場。 工業機器人推動智能製造 在澳門科學技術發展基金首屆重點研發專項資 助計劃支持下，徐教授的團隊也在開發擁有三維 視覺感知和柔順力控的新一代工業機器人，能夠 在可變化的環境與人類緊密合作。所謂柔順力 控，即是機器人能夠因應外力影響（例如是機器 臂活動時被人碰到一下）靈活調整自身力度，確 保完成任務。 徐教授說，新的機器人將會比同類產品更靈活、 更智能，能完成更複雜的工作。這個項目的目標 是提升機器人智能作業系統的性能，將以零件裝 配、汽車打磨、飛機維護等任務作為示範用途，相 信成果會有利推進澳門經濟適度多元化發展。 消毒機器人 A disinfection robot 機器人微夾鉗系統 Robotic microgrippers
COVER STORY • 封面專題 2021 UMAGAZINE 24 • 澳大新語 18 to adjust its force flexibly in response to external forces, such as a person pushing a robotic arm, to continue its task. Prof Xu says that his new robot will be more flexible and intelligent than its counterparts in use, and will be capable of performing more complicated tasks. Moreover, his team will use the new robot to assemble some parts of devices, polish cars, and maintain aircraft. The application of research outcomes will be conducive to the diversification of the local economy, Prof Xu adds. Microrobots for Biomedical Engineering Prof Xu’s team has also developed an intelligent micromanipulation robot system that can automatically control microinjectors and microgrippers at the micro to nano scale for cell operations such as gene editing. This system can make microinjection more stable and reliable by ensuring a high cell survival rate. He says the new system can meet the ever‑growing demand for the cell operations in biomedical engineering. Furthermore, the team has also designed a new algorithm for precise hybrid motion and force control. This intelligent algorithm will help the robot perform micromanipulation more accurately and quickly. The team’s project, ‘Development and Industrial Application of Robotic Microgripper System’, is among the first batch of scientific and technological innovation projects at the University of Macau‑Huafa Group Joint Laboratory. They are also supported by the Zhuhai UM Science & Technology Research Institute to transform their research results into marketasle products. Over the past decade, Prof Xu has served on editorial boards of several top international journals and has been awarded the Macao Science and Technology Award multiple times, and his team is still striving for further advances. Prof Xu says, ‘We’re also developing vascular robots, aerial robots, and many other intelligent robots, in order to improve people's lives through technological innovations. I am confident that intelligent robots will be closely integrated with people to better serve society.’ Researchers around the world have been exploring new technologies to develop intelligent robots, a phenomenon driven by advances in artificial intelligence and its ever‑growing demand from industries. At the University of Macau (UM), Xu Qingsong, a professor in the Department of Electromechanical Engineering, is leading a team to build innovative intelligent robots to improve people's lives and serve society. Disinfection Robots for Pandemic Response Prof Xu began to design a disinfection robot called ‘Smart Cleaner’ during the early stage of the COVID‑19 pandemic. The project was funded by the Macao Science and Technology Development Fund (FDCT), and the Institute of Development and Quality provided support by monitoring air quality indicators. The robot not only can reduce the need for human cleaners and the risk of virus transmission, but also can ensure that disinfectant is evenly sprayed. ‘We did everything ourselves, from the development of hardware to programming, design, and assembly. That’s why we can keep the cost low, which is important for promoting the robot,’ Prof Xu says. According to Prof Xu, UM has applied for a patent for the robot technology. The university is also talking with companies for patent transfer, in order to put the robots on the market in the Greater Bay Area and beyond. Industrial Robots for Smart Manufacturing The team is also developing a next‑generation industrial robot with three‑dimensional visual perception and compliant force control, with the support of the FDCT’s first Macao Funding Scheme for Key R&D Projects. Compliant force control refers to the ability of a robot 徐青松教授 Prof Xu Qingsong 掃二維碼 觀看訪談片段 Scan the QR code to watch the interview
澳大新語 • 2021 UMAGAZINE 24 19 封面專題 • COVER STORY 徐青松教授團隊研發的第一代「消毒智多星」在澳門鏡湖醫院噴灑消毒液 A first-generation Smart Cleaner, developed by Prof Xu Qingsong’s team, sprays disinfectant in Kiang Wu Hospital in Macao.
COVER STORY • 封面專題 2021 UMAGAZINE 24 • 澳大新語 20 文/葉浩男‧圖/何杰平、部分由受訪者提供 Chinese & English / Davis Ip ‧ Photo / Jack Ho, with some provided by the interviewee The AI Image Forgery Detective 智能偵測圖像篡改 眼看未為真 周教授是澳大人工智能與機器人研究中心代主 任，也是智慧城市物聯網國家重點實驗室（澳門 大學）的成員。他說：「許多被篡改的圖像不但 肉眼無法察覺，連電腦程式也偵測不到。」 科技令圖像篡改變得輕而易舉、電子文件真偽 難測。澳門大學電腦及資訊科學系副教授周建濤 的團隊憑著一款先進算法，在一場圖像篡改偵 測國際比賽中擊敗1 , 500多支隊伍，目前與阿 里巴巴合作開發更強大的偵測工具。 周建濤教授團隊借助澳大智能超算中心的超級電腦訓練用於圖像篡改偵測的深度學習模型 Prof Zhou Jiantao’s team uses the Super Intelligent Computing Centre at UM to train their deep learning models for image forgery detection
澳大新語 • 2021 UMAGAZINE 24 21 封面專題 • COVER STORY 圖像被篡改的位置，準確度遠超對手。這款算法特 點是採用了一個多網絡架構的空間通道感知模組， 能夠準確地提取圖像特徵。 產學合作 憑著矚目的表現，周教授的團隊獲阿里巴巴贊 助加強算法。周教授說，網上購物平台每日都要 驗證大量網店的牌照，確保賣家都是合資格的商 戶。面對高解像度的圖像時，現有算法一般可以 準確偵測出經篡改的圖像，但處理低解像度圖 像，例如是經社交媒體或通訊軟件壓縮過的圖 像，往往束手無策。 研究團隊正在參與「阿里巴巴創新研究計劃」，開展 為期一年的「抗媒體傳輸的高魯棒偽造圖像檢測與 定位研究」，旨在設計更高效的偵測算法，即使目標 圖像曾被不同媒介壓縮、調整大小、過濾或添加噪 聲，仍能找出破綻。 周教授說，這項目是澳大在該領域與大型科技企業 首次合作，有助他的團隊累積產學研合作經驗：「我 們正在運用研究成果解決現實世界的商業問題，進 展令人鼓舞。」 圖像篡改偵測算法有如專業偵探，察覺到旁人不為 意的蛛絲馬跡。這些算法通常會分析圖像的噪聲分 佈和其他特徵，尋找線索。「如果一幅圖像未經篡 改，整幅圖的噪聲分佈通常會保持一致。」 2019年起，研究團隊獲澳門科學技術發展基金資 助，開展一項關於準確分析噪聲和提取圖像特徵的 研究項目，其成果有助開發偵測算法，令篡改過的 圖像無所遁形。 探微知著 2021年初，周教授的團隊參加由清華大學和阿里 巴巴合辦的「安全AI挑戰者賽（第五期）」，在「篡改 賽道」勇奪冠軍，也在「檢測賽道」獲得季軍。在「篡 改賽道」，團隊修改了20張證件類圖像上的資訊， 例如身份證上的姓名和出生日期。他們會分析圖像 中真實部分的噪聲，同時參考被篡改部分的背景細 節，最後在被篡改部分添加一層自適應噪聲，用來 躲避人工智能工具的偵測。他說：「我們是1,534支 參賽團隊中最成功的圖像篡改者。」 比賽期間，他們也訓練出一款新的偵測算法。它經 過深度學習數以萬計的圖像，不出半秒就能偵測出 研究人員將SE-Block結構加入到神經網絡，提升其從圖像擷取資訊的效能。 圖為SE Block的結構。 The structure of a Squeeze-and-Excitation Block. Such blocks are plugged into the neural network to improve its performance in extracting information from images. 周建濤教授（右）的團隊在由清華大學和阿里巴巴合辦的 「安全AI挑戰者賽（第五期）」獲獎 A team led by Prof Zhou Jiantao (right) has won prizes at the Security AI Challenger Contest (Season 5), an algorithm competition organised by Tsinghua University and Alibaba.
COVER STORY • 封面專題 2021 UMAGAZINE 24 • 澳大新語 22 has never been altered, there should be a consistent noise pattern throughout it,’ Prof Zhou says. Since 2019, his team has developed new ways to accurately examine noise and extract features from images, in a project supported by the Macao Science and Technology Development Fund (FDCT). This project has inspired new algorithms for detecting forged images. The Truth in the Details In early 2021, Prof Zhou’s team won the championship in the image forgery track, and the third prize in the detection track, at the Security AI Challenger Contest (Season 5). At this algorithm competition organised by Tsinghua University and Alibaba, the team altered 20 images, such as changing names and dates on ID cards, and added a special adaptive noise. ‘We generated the noise after analysing background details and noise in the authentic parts, in order to hide the forgery traces from AI detection tools. We turned out to be the best forger among the 1,534 teams,’ Prof Zhou says. It has never been easier to create forged images, which are now very convincing. At the University of Macau (UM), the team of Associate Professor Zhou Jiantao in the Department of Computer and Information Science has developed a cutting‑edge algorithm to detect such images. After defeating over 1,500 rivals at an international competition, they have continued to improve the algorithm under a project with the e‑commerce giant Alibaba. Seeing Is Not Believing Prof Zhou is the interim head of UM’s Centre for Artificial Intelligence and Robotics, as well as a member of the university’s State Key Laboratory of Internet of Things for Smart City. ‘In many images, the forged parts are undetectable to the naked eye, sometimes even to computer programmes,’ he says. Like a human detective, a good image forgery detection tool has to recognise details that others might overlook. In digital images, such details include noise and other features. ‘For instance, if an image 周建濤教授運用團隊開發的算法，將一張食品經營許可證被篡改的部分偵測出來。 The algorithm developed by Prof Zhou Jiantaoʼs team can detect the forged parts of an image of a food business licence
澳大新語 • 2021 UMAGAZINE 24 23 封面專題 • COVER STORY Under the Alibaba Innovative Research Programme, the tech company has sponsored Prof Zhou’s team to conduct a one‑year project titled ‘Research on Highly Robust Methods for Detecting and Locating Forgery in Images Transmitted Through Media’. It aims to develop an algorithm that can accurately detect forged parts of images even if they have been compressed by different media, resized, filtered, or contain added noise. ‘This is UM’s first collaboration in this field with a big tech company,’ Prof Zhou says, adding that his team has gained valuable experience in business‑university collaboration. ‘We’re applying our expertise to meet real‑world business challenges, and have seen encouraging progress.’ For the same competition, Prof Zhou’s team also trained an algorithm that, after learning from tens of thousands of images, can detect forged areas in less than half a second. It outperformed most of its competitors largely due to a multi‑network architecture that integrates spatial channel perception modules, which gives it an exceptional power for extracting features from images. A Business Partnership The team’s performance has led to its collaboration with Alibaba to make the algorithm more robust. Every day, online marketplaces like Alibaba’s need to verify countless business licenses to make sure they are dealing with legitimate sellers. Existing algorithms perform reasonably well in detecting forgery in high‑resolution images, but they are not very effective with low‑quality, smaller images, which have usually already been compressed by messaging applications or social media platforms. 周建濤教授團隊的算法可在半秒內偵測出圖像被篡改的位置 Prof Zhou Jiantaoʼs team has developed an algorithm that can detect forged areas in an image in less than half a second 周建濤教授團隊運用澳大的智能超算中心來訓練算法。該 中心提供多個GPU計算平台，可以執行深度學習任務和作 為虛擬數據中心 。 Prof Zhou Jiantaoʼs team uses the Super Intelligent Computing Centre at UM to train their algorithms. The centre hosts GPU computing platforms which can run deep learning tasks and serve as virtual data centres.
COVER STORY • 封面專題 2021 UMAGAZINE 24 • 澳大新語 24 文/葉浩男、校園記者林程峰‧圖/何杰平、部分由受訪者提供 Chinese & English / Davis Ip, UM Reporter Lam Cheng Fong ‧ Photo / Jack Ho, with some provided by the interviewee 社交媒體數據驅動智慧旅遊 Social Data Drives Smart Tourism 鞏志國教授的團隊開發的手機程式能提供個人化行程推薦 Prof Gong Zhiguo’s team has developed a mobile application that can plan personalised itineraries
澳大新語 • 2021 UMAGAZINE 24 25 封面專題 • COVER STORY 店、餐廳和博物館等旅遊場所都能採用這些技 術，了解旅客需要，改善服務和拓展客源。」他的 團隊也正研究用社交媒體數據，評估各區不同 時段的旅客密度，為政府規劃交通及旅遊設施 提供參考。 改良算法推動智慧旅遊 為方便市民和旅客出行、避開塞車，研究團隊 將會推出另一款手機程式，數據來自4 0 多個 路口的攝錄鏡頭。程式能在地圖上即時顯示道 路的擠塞程度和預測將來的路況，主要運用兩 種機器學習方法，分別是能從圖像識別車輛數 量和類型的「卷積神經網絡」，以及用來分析 相關的時序數據的「循環神經網絡」。他們更 在開發另一款運用機器學習模型的程式，預測 巴士到站時間。 鞏教授說，他的團隊還在處理一些技術挑戰， 其中一個是各大社交媒體開放數據的程度不 一、數據格式各異，需要整合不同平台產生的 數據。分析社交媒體數據時也會遇上語義分析 的難題，需要令電腦更準確地理解有多種含義 的單詞和句子：「我們的團隊將會繼續改進機 器學習模型，在澳門為旅客創造更優質和智能 的旅程服務。」 社交媒體每天產生大量時空數據，充分利用這 些大數據可以推動旅遊業智能化。為了善用這 些數據，澳門大學的研究人員正在開發一系列 先進算法，藉此推動澳門成為智慧型世界旅遊 休閒中心。 數據助旅客規劃行程 在智慧城市物聯網國家重點實驗室（澳門大學）， 數據挖掘專家、電腦及資訊科學系主任鞏志國 教授正在與研究生研究「社交媒體數據流的在 綫事件檢測與智能分層聚類技術」，其中一環 是開發供旅客和業界使用的手機程式。項目在 2019年起獲澳門科學技術發展基金資助。 鞏教授的團隊設計了一款手機程式，能夠推薦 個人化的澳門行程，原理是分析旅客在社交媒 體留下的「遊覽軌跡」，主要是他們在澳門到訪 過的地點、時間和評論。鞏教授相信，提供更個 人化的旅遊體驗有助延長旅客在澳遊覽的時間。 「對初次來澳的旅客，我們也運用了遷移學習 技術，通過分析他們在原居地的遊覽軌跡獲取 其旅遊喜好，在他們抵達前就能推薦景點。」 與此同時，鞏教授的團隊開發了一些新算法分 析社交媒體數據，能夠評估旅客心情、找出熱門 的名勝和活動，以及識別和預測突發事件。「酒 鞏志國教授 Prof Gong Zhiguo 電腦程式顯示旅客的位置和心情 A computer programme shows the locations of tourists and their sentiments
COVER STORY • 封面專題 2021 UMAGAZINE 24 • 澳大新語 26 Among the many products resulting from their research is a mobile application that can recommend personalised itineraries for users. It works by analysing the ‘travel trajectory’ of the users on social media, which is largely a list of places they have travelled in Macao tagged with visiting times and comments. Prof Gong believes that providing more personalised travel experiences will encourage tourists to stay longer in Macao. ‘Our app also uses transfer learning algorithms to analyse the travel preferences of visitors in their hometowns, so that it can recommend travel options once they arrive at Macao,’ he says The UM research team has also developed new algorithms that use social media data to evaluate the tourists’ sentiments to identify their favourite places and activities, and to detect and predict emergent events. ‘New technology allows hospitality venues such as hotels, restaurants, The massive data we create every day on social media holds the key to transforming the travel industry. At the University of Macau (UM), researchers are designing advanced algorithms which can help Macao become a smart world centre of tourism and leisure. Tailor‑made Tours for Everyone Prof Gong Zhiguo is a data mining expert and head of UM’s Department of Computer and Information Science. At the State Key Laboratory of Internet of Things for Smart City (University of Macau), Prof Gong and his students are designing mobile applications for tourists and the tourism sector. Their project, titled ‘A Hierarchical Categorisation Model for Online Events Discovery from Social Media Data’, has received support from the Macao Science and Technology Development Fund since 2019. 澳大正在開發一系列先進算法，推動澳門成為智慧型世界旅遊休閒中心。 UM researchers are designing advanced algorithms which can help Macao become a smart world centre of tourism and leisure
RkJQdWJsaXNoZXIy MTQ1NDU2Ng==