Digital Twin Tokyo goes live! Embodied Intelligence Zero Sample Migration Real World, Shared Swarm Thinking

[New Wisdom Introduction] After empowering the fields of art design and text writing, AI has also injected new vigour into urban planning and real-time monitoring of urban dynamics.
Tokyo's high-resolution point cloud 3D digital twin model is now publicly available! Anyone can download it for free.

Project address: https://github.com/tokyo-digitaltwin

The scale of the digital twin model is huge, and its depiction of Tokyo is incredibly detailed - its absolute positional accuracy is within about 10cm.

Zebra crossing on the road

Jim Fan, NVIDIA's Senior Research Scientist, said, ‘It's a given that more and more cities, houses, and factories will be imported into simulated environments’.

In the future, robots will not be trained in isolation. They will be simulated as a ‘fleet of steel’ in a real-time graphics engine, scaled up with a huge cluster to generate the next trillions in high-quality training data.
In other words, digitising the physical environment and importing it into the virtual simulation world will greatly accelerate the development of robotics.
By training in high-precision simulated environments, robots can acquire rich training data and learn quickly in complex scenarios.
This approach will promote the smooth migration of robots from the virtual world to the real world, and enhance their efficiency and intelligence in practical applications.

What is an urban digital twin?
Simply put, a digital twin is a reproduction of an entity in physical space moved to cyberspace.
Once this reproduction reaches a certain level of accuracy, one can simulate behaviours and monitor operations by using real-time data sent from sensors on the physical objects, enabling the construction and exploitation of ‘twin’ cities.
Digital twinning of cities is actually the reproduction of infrastructure such as buildings and roads, economic activities, human flows, and other elements like ‘twins’ in cyberspace based on perceived urban data.

In other words, as in the continuous loop like the one in the figure below, it is possible to perform advanced analyses and simulations in cyberspace based on real-time data obtained from various areas of activity in the physical space and to feed the results back to the physical space in an interactive form at high speed.

According to Werner Kritzinger et al. there are three phases of digital twinning.
The first stage is known as digital modelling, where the interconversion between an existing physical object and the virtual space in which it is represented is a manually performed state. In the case of the digital twin of the city, it is the state in which the historical data already available (e.g., reproduction of traffic flow data, etc.) is reproduced on the 3D city.
The second stage is known as digital shading, a state of unidirectional data flow that supports the automatic conversion of physical objects into digital objects.
The third stage is the digital twin, where physical and digital objects, achieve a complete fusion in both directions, where each object changes and is automatically reflected in the other. In the digital twin of the city, it is the state of two-way data exchange in the real physical space and cyberspace.

For the digital twin of the city, the main values are:
- Real-time data acquisition linked to reality: collecting various data in real time using advanced sensor technologies and communication technologies
- Analyses and simulations using 3D space: advanced analyses and simulations such as experiments in a space that reproduces reality
- Feedback to reality: real-time feedback of results to real space for decision-making, system control, etc.
For example, the application of digital twins and AI prediction engines to traffic congestion analysis and regulation in New South Wales can significantly reduce the cost of social benefits lost due to traffic congestion.

Digital Twin vs Analogue
A city digital twin is a digital replica of a city built to reflect the state of the physical world in real time.
It constantly changes as the city runs in real time, like a ‘shadow’ that lives and breathes with the real city. The model not only reflects the current state of the city, but is also able to react quickly to unexpected situations in reality.
In contrast, city simulation is a relatively static model based on assumptions and preset rules.
It is a distillation and summary of past city operation data, and although it can reflect certain patterns, it is less accurate than the real-time data of the digital twin. City simulation is more like projecting possible future conditions based on the city's past ‘medical history’.
Overall, digital twins and simulations are both simulations based on virtual models, but there are some key differences.
Simulation is typically used for design and, in some cases, offline optimisation. Designers input changes into the simulation to observe what-if scenarios. Digital twins, on the other hand, are complex virtual environments that people can interact with and update in real time. They are much larger and have a wider range of applications.
In automotive simulations, for example, new drivers can get an immersive training experience, learn the operation of various car parts, and face different real-life scenarios while driving virtually. However, these scenarios are not associated with the actual physical car.
The digital twin of the car, on the other hand, is associated with the physical vehicle and learns all the information about the actual car, such as important performance statistics, parts that have been replaced in the past, potential problems observed by sensors, previous maintenance records, etc.

The Three Pillars of the Urban Digital Twin
The three pillars of Urban Digital Twin are defined as ‘Data Maintenance’, ‘Data Visualisation’ and ‘Data Analytics’.

- Data Maintenance
Maintenance and aggregation of geospatial data information processed on the digital twin such as 3D digital maps, point cloud data, GIS data, etc., which includes data collection, data storage and management.

- Data Visualisation
Through visualisation systems such as 3D viewers, complex data is transformed into intuitive graphs, charts and 3D models, enabling city managers to see a ‘digital portrait’ of the entire city.
For example, heat maps are used to show the population density in different areas of the city, dynamic flow diagrams are used to represent the flow of traffic and congestion, and 3D models are used to simulate the distribution of energy consumption in buildings.
These visualisations allow city managers to see the state of the city's operation at a glance, as if they have a pair of ‘eyes’ that can see through the city.

- Data Analysis
Through various application simulators, the data on the digital twin is analysed and used for measures.
Taking urban public transport as an example, by analysing data such as passengers‘ travel time, travel routes and travel frequency, it is possible to adjust bus routes, optimise the scheduling of buses, improve the operational efficiency of public transport, and reduce passengers’ waiting time and travel costs.
In terms of energy management, analysing the energy consumption data of buildings, accurate energy saving plans can be formulated to reduce the city's energy consumption and achieve sustainable urban development.

Data maintenance, data visualisation and data analysis

Tokyo Digital Twin
The Tokyo Digital Twin project is planned to be realised by 2030, and the related beta version has already been launched.

Link to experience: https://3dview.tokyo-digitaltwin.metro.tokyo.lg.jp/? _ga=2.180157796.1370176531.1735090358-57006728.1735090358