Research

H2O revolves around four research programmes, to bring AI at the core of future smart and sustainable cities. From the interfaces for Human-AI interactions to the social implications of AI, H2O will focus on designing the future of connected cities, combining technologies such as AI, 5G, AR, and Data Science.

Research Programme 1: Immersive Interfaces for Human-AI Interactions

With the explosion of the IoT, AI will become necessary to analyse and interpret the data and to handle large fleets of devices. AR can provide an immersive user interface for Human-Data and Human-AI interactions. Both technologies heavily rely on machine learning algorithms, which require heavy computational power combined with large databases supported by the network infrastructure. To ensure real-time analysis and visualization, a combination of cloud and edge systems brings the computations to the user more efficiently.

H2O aims at making Hong Kong the first AR city in the world. An AR city does not only rely on augmented reality but exploits multiple technologies (IoT, Big Data), brought together by a strong communication medium such as 5G. Through innovative interfaces, AR enables the visualization of large amounts of data. AI can enhance techniques such as speech interaction, navigation, and human-computer interaction to improve the AR user experience. In particular, H2O will explore the following topics: (1) Cloud and Edge-based AI platforms for AR, (2) AR content creation and modelling (3) AI-enhanced AR.

Research Programme 2: AI for sustainability

Urban sprawl is fast and inevitable. With this rapid expansion, sustainability becomes a challenging issue. Most infrastructures are not adapted to the current population, which causes various problems including traffic congestion, increased pollutants emissions, escalation of waste production and insufficient recycling facilities, meanwhile increasing poverty, and inflation of housing prices. To circumvent these challenges, more and more sensors have been deployed around cities, ranging from traffic cameras to air quality monitors. These sensors constantly monitor the state of the city in close to real-time. However, the amount of data generated is difficult to interpret at various scales.

The usage of AI within the city environment allows us to exploit the increasing amount of data generated in real-time. AI can also interpret data at various scales, both geographical and temporal, and infer the city-wide long-term consequences of decisions. Finally, AI allows us to study how users interact with the cities and the environment. This data can also be used to improve world models by incorporating user behaviour. H2O will consider the following themes for environmental AI: (1) AI for disaster response, (2) AI for sustainable cities, (3) AI for intelligent transportation, (4) Smart homes for Smart cities.

Research Programme 3: AI for Communications and Networking

AI is pervasive by nature. However, it is also computation-hungry. In order to accommodate this paradox, H2O aims at developing new architectures relying on 5G and potential 6G. Planned to be available within 2020, 5G will allow a wide range of intelligent applications to be developed. Hong Kong would greatly benefit from being one of the first regions to deploy commercial 5G and display its capabilities through bleeding edge applications.

Data-driven intelligence lets machines learn for themselves by providing them access to large amounts of data. Future networks will be flatter and edge devices will play a vital role in mobile networks, but also raise specific challenges. More specifically, how to: store and analyse data streams from diverse edge devices efficiently, synchronize AI modules between cloud and edge; guarantee relevant and legitimate data are processed transparently. The next generation of mobile communications (6G) will rely even more on Artificial Intelligence and Data Science. In particular, the network will be so complex that it will require AI to take control and adapt the conditions faster and more efficiently than a human being can ever do. The topics covered within this program are: (1) Bringing AI closer to users, (2) 5G driven applications and AI, and (3) AI-driven 6G design.

Research Programme 4: AI for Humanity and Society

We already previously evoked AI for city planning. This concept can be extended to the social aspects of a city and employs AI for improving human life. H2O aims at building upon this existing work, and develop novel strategies for human-oriented, AI-assisted city planning.

AI for humanities does not limit itself to external factors. AI can directly involve into human-centred topics such as education or healthcare. One of the specific aspects of these topics is the constant presence of humans in the decision process. One final example of using AI for society is related to privacy. AI systems open many ethical questions such as which organisms will regulate it, what we define as an unbiased AI, or to what extent can AI influence users' decisions. Ensuring privacy bounds not only on the data gathered but also on the AI influence aspect is a key issue for the next-gen internet architecture. H2O will consider the following research points in combining AI and humanities: (1) AI for education, (2) AI for health and well-being, (3) Secure AI for social good, and (4) Digital humanity.