Scholarlink

<p>The digitization of planning processes in urban development is gaining new momentum thanks to the possibilities offered by artificial intelligence (AI). One very complex process is the integration of legally binding existing urban land-use plans into the XPlanung data standard. Associated delays in the introduction of digital procedures for building permits lead to lengthy procedures for urgent urban development needs.</p><p>The article shows how AI-supported processes can accelerate the digitization of planning documents. Based on selected development plans of varying ages it is clear which urban land-use planning indicators can be reliably mapped and transferred to current technical schemata. The results show that multimodal large language models (MLLMs) are already capable of interpreting textual and machine-readable building regulations in plans. In order to recognize unstructured content (e.g., overlaps of text and plan symbols) and georeferenced plan content with topological spatial references (e.g., setback lines), MLLMs, however, still require additional specific training data with planning domain knowledge.</p><p>Corresponding improvements are to be expected in the near future, which will significantly benefit the efficiency of digitizing planning processes for urban development. Practical implications arise primarily for infill development. Here, there is an increasing number of older development plans that could be processed using AI-supported digitization. This development is an important prerequisite for the creation of urban digital twins and their fields of application for infill urban development. New standards provide the relevant guidance for practical transfer.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Urbane Digitale Zwillinge: Anwendungsperspektiven für die Stadtentwicklung durch die KI-gestützte Digitalisierung von Bauleitplänen

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月25日

查看全文
  • Stefan Fina,
  • Michael Schwarz,
  • Nayab Bushra,
  • Tobias Maile

摘要

The digitization of planning processes in urban development is gaining new momentum thanks to the possibilities offered by artificial intelligence (AI). One very complex process is the integration of legally binding existing urban land-use plans into the XPlanung data standard. Associated delays in the introduction of digital procedures for building permits lead to lengthy procedures for urgent urban development needs.

The article shows how AI-supported processes can accelerate the digitization of planning documents. Based on selected development plans of varying ages it is clear which urban land-use planning indicators can be reliably mapped and transferred to current technical schemata. The results show that multimodal large language models (MLLMs) are already capable of interpreting textual and machine-readable building regulations in plans. In order to recognize unstructured content (e.g., overlaps of text and plan symbols) and georeferenced plan content with topological spatial references (e.g., setback lines), MLLMs, however, still require additional specific training data with planning domain knowledge.

Corresponding improvements are to be expected in the near future, which will significantly benefit the efficiency of digitizing planning processes for urban development. Practical implications arise primarily for infill development. Here, there is an increasing number of older development plans that could be processed using AI-supported digitization. This development is an important prerequisite for the creation of urban digital twins and their fields of application for infill urban development. New standards provide the relevant guidance for practical transfer.