UTA members Keio University and Nagoya University recently organised a 1 day workshop in Tokyo to discuss the latest technology trends affecting cities.
Other participants included Japanese Telecom company NTTE, The National Institue of Informatics NII, SMEs such as Augmented reality company Kadinshe and IoT company Green Blue as well as a number of academics including representatives from Cambridge University in the UK.
Keynote – Levent Gurgen Kentyou and UTA
Kentyou founder, Levent Gurgen gave the opening keynote, in which he started with a brief history of UTA, explaining how the EU-JP research projects highlighted the need for increased collaboration between cities, academia and industry which was the genesis of UTA.
At the technology level, the collaborative projects highlighted the need for an open, flexible, city data platform which was a key element of several of the EU-JP projects and which Levent has gone on to commercialize via his company, Kentyou.
Dr. Gurgen finished by highlightng a number of recent projects that leveraged UTA’s matchmaking to develop a world class team to meet the project brief, and exploited the Kentyou city platform. These included work on Smart Intersections for traffic flow and safety in New York and Istanbul, as well as the Grenoble based climate centric Climaborough project addressing Urban Heat Islands in cities.
Guest Speakers
Nick Lane – Cambridge University, UK. Machine learning and the data center – a dangerous dead end
Prof. Lane addressed the future direction of Machine Learning, and in particular focused on the potential problems of the current trend towards centralised data center based ML highlighting 4 key issues:
- User privacy and control – a centralised approach forces compromises in user privacy and control.
- Brittleness of systems – edge devices require access to the central data centre to operate, providing a single point of failure.
- Data centres have significant energy and CO2 footprints – because its all done from scratch
- Bias in the models – a fundamental fact of the centralised model design.
His proposed solution is a decentralised and collaborative approach which supports federated learning – addressing most of the issues highlighted above. To prove the value of this approach his team has developed an open source tool for federated learning called Flower which has been spun out as a new startup Flower.ai
Cecilia Mascolo – Cambridge University, UK. Using wearable data for behaviour analysis
Prof. Mascola gave an overview of some of her research over the past 10 years focusing on the value and use of wearable data. She highlighted the different devices that have been used by the research community and how different data sets can be used to discover intersting insights both about users but also their surroundings. Some examples included:
- Data analysis in the city – using 4square checkins her team explored how cities are used in different ways at different times
- Continuing the long tradition at Cambridge of using smart tags in the lab and work environment her teams has explored tags as a tool to understand office behaviour. She highlighted how data can be used to better design workspaces based on a recent move by the team to a new building.
- Cecilia finished up with a discussion of recent work on earbuds – highlighting design issues and their ability to be used for a variety of tasks including gesture recognition, gait and health monitoring.
Partner updates
Following on from the Keynote and guest speaker sessions, the workshop continued with a set of presentations from UTA members, these included:
Fuyuki Ishikawa – NII
Prof. Ishikawa gave a quick update on some of his activities since the BigClouT project where his team focused on using AI to model activity in the city. His focus is now on dependability – “dependability of AI & AI for dependability”
Jin Nakazawa – Keio University
Prof. Nakazawa highlighted recent work in his lab including work on AI and QoS to understand both functional but also non-functional requirements and the recent JST Crest project that is trying to better understand technologies and infrastructure for behaviour change. More details are to be found here
Soko Aoki – Kadinshe
Dr. Aoki highlighted the recent work of his company – Kadinche – which has been developing VR, AR and mixed reality technologies for a number of years. He briefly showed work with a mixed reality app for bus travel through a regional park in Japan as well as VR of a virtual entertainment experience on a Ferris wheel.
Takuro Yonezawa – Nagoya University
Dr. Yonezawa discussed recent work on digital twins and the metaverse. He highlighted research questions around new realities relating to mixed reality enviornments and finished with a brief demo of his labs’ Metapo mixed reality social robot.
Rodger Lea – Urban Technology Alliance
UTA’s CTO, Dr Rodger Lea gave a quick update on the activities of UTA highlighting the role of UTA in building teams to address city problems via its matchmaking service and recent projects including the Climaborough project and the EU funded Cognifog project. He finished with a brief introduction of new UTA members including Latitudo40, Realsim, Crowdscan and Caspera Labs
Discussion – what key technology trends should we be considering
The workshop finished with a lively discussion on technology trends that had been highlighted during the day or that had bubbled up in the discussions. These were captured under 8 key areas.
GenerativeAI and Cities: There was general agreement that GenAI will have an impact on cities and in particular how city staff manage their daily activity. It was noted that UTA has already posted about this issue. Yokosuka city in japan was highlighted as a leading city in this area and Keio discussed a recent workshop they ran on GenAI technologies. It was felt that UTA could serve a useful role by providing training and information on GenAI for its city members.
3D digital twins. There has been much discussion on DigitalTwin technology – this ranged from the work of Kentyou with it’s 2D dashboards, through to the Kadinche work on full immersive 3D. Additionally there seemed to be overlap in the work on dependable AI that FI presented and the use of AI in DigitalTwins. Participants felt that the merging of AI and DigitalTwin was an area worth further exploration.
Social and Health. Several of the projects reported by workshop members touched on issues of health and social wellbeing and the need for technology to consider and in some cases support health and wellbeing in cities. This ranged from the work on Urban Heat Island through to mixed reality tools to engage citizens. It was noted that this area need significant further work as currently technology is resulting in social fragmentation an disharmony.
Non-urban simulation/visualization. Several participants discussed AR/VR’s rle in smart cities and highlighted that the main focus is DigitalTwins. However, the technology has broader application, for example some of the work that Kadinche has been doing on AR/VR and simulation including projects in Soleil park on augmenting the parks bus ride with virtual animals. These type of non-urban environment are also worth considering as UTA broadens its focus from city to mixed urban/rural areas.
Federated Learning and distributed systems. The presentation by Nic Laneon federated/distributed learning touches on a number of distributed systems work areas that have been studied over the years. This trend towards decentralised infrastructure is important for cities and is being explored by the Cognifog project as well as others under the EU’s Cloud Edge IoT continuum
Augmenting peoples reality. It was highlighted that the idea of different types of reality becoming more acceptable in urban settings ranging from VR through to AR (and other forms of reality). Several participants felt that there was an interesting spectrum of choices in Urban Mixed Reality and wanted to explore this further to help cities exploit the technology for social and citizen improvement.
Emergency Management tools. It was observed that a number of the presentation touched on tools (through visualization and simulation) that were targeted at, or could be applied to Emergency situations. It was a use-case in the Cognifog Paris, and a significant issue in Japanese urban planning. Further work was needed to exploit some of the technologies discussed at the workshop in aid of urban emergency planning.
Effective info delivery. A common thread through several of the projects was the notion of effective information delivery. Different circumstances require that different means of communication are used, for example, in emergency situations how best to communicate information. Equally, if participants are immersed in a VR or AR scenario how best to communicate info.
Conclusions
The workshop covered a number of interesting areas and highlighted ongoing activities of UTA members and associates. Although there was a broad range of diverse topics discussed, the 8 key areas highlighted above were considered the most significant and of considerable interest to cities as they navigate the transition to smarter cities.
