New headline: Artificial intelligence can create urban designs beyond human capabilities, study shows

Imagine living in a cool, green city full of parks, running through sidewalks, bike lanes and buses that quickly get people to shops, schools and service centres. Just a few minutes

This breezy dream is the epitome of urban planning, embodied in the concept of 15-minute cities, where all basic needs and services are within 15 minutes, improving public health and reducing vehicle emissions.

Now, artificial intelligence can help city planners realize this vision faster, with a new study from researchers at Tsinghua University showing how machine learning can generate more efficient spatial layouts in a short time than humans can design.

Scientists at Tsinghua University hope to find new solutions to improve our cities, which are rapidly becoming crowded and concrete.

They developed an artificial intelligence system to solve the most tedious computing tasks in urban planning. The study found that the urban plans generated by the system were about 50% better than human designs on three indicators: accessibility of services and green spaces, and traffic levels

Starting small, the research team asked their model to design urban areas that were only a few square kilometers in size (about 3 × 3 blocks).

After two days of training and using several neural networks, the AI ​​system searches for ideal road layouts and land uses to fit the 15-minute concept of cities, local planning policies and needs.

While the researchers’ AI model has some capabilities that could be extended to planning larger urban areas, designing entire cities would be more complex. Researchers estimate that drafting a neighborhood of 4×4 blocks would require twice as many planning decisions as a 3×3 block.

But automating even a few steps in the planning process can save significant time: AI models calculate certain tasks in seconds that would take a human planner 50 to 100 minutes to complete.

Researchers note that automating the most time-consuming urban planning tasks could allow planners to focus on more challenging or human-centered tasks, such as public engagement and aesthetics

Rather than AI replacing humans, the research team envisions their AI systems serving as "assistants" to urban planners, where they can generate conceptual designs optimized by algorithms and reviewed, tweaked and reviewed by human experts based on community feedback. Evaluate. What it's like: Instead of AI replacing humans, the team envisions their AI systems acting as "assistants" to urban planners, generating algorithmically optimized conceptual designs that are reviewed by human experts based on community feedback. , adjust and evaluate

Paolo Santi, a research scientist at the Massachusetts Institute of Technology (MIT), noted in a commentary on the study that good design is the key to success

Urban planning, he writes, “is not just about allocating space for buildings, parks, and functions, but about designing a place where urban communities live, work, interact, and hopefully thrive in the long term.”

The research team compared their AI workflow to human design and found that the collaborative process increased usage of essential services and parks by 12% and 5% respectively

The researchers also surveyed 100 urban designers who did not know whether the plans they were asked to choose were generated by human planners or artificial intelligence. As a result, some AI space designs won significantly more votes, but for other plans, survey participants showed no clear preference.

The real test, of course, will be in building communities under these plans, measured against reducing noise, heat and pollution, and improving public health, while delivering on the promise of better urban planning

This research was published in the journal Nature Computational Science

If friends like it, please follow "Know What's New"!

The above is the detailed content of New headline: Artificial intelligence can create urban designs beyond human capabilities, study shows. For more information, please follow other related articles on the PHP Chinese website!