Abstract:The competition for limited urban spaces due to the growing demand for both living space and green areas causes conflicts in urban development. This paper aims to support urban planners with a systematic multi-criteria decision making process and a supporting tool-chain. The process uses multi-objective optimization (MOO) to quantify trade-offs, allowing planners to gain a broad perspective on optimal solutions. The generic process builds an urban simulation model, applies MOO, explores the results in terms of multi-criteria trade-offs, and guides the decision making of urban planners. To enable this process, we present a tool-chain for applying a Gaussian Process Regression based MOO algorithm to a computationally expensive urban simulation model. The tool-chain allows to identify Pareto-optimal solutions and their properties with reasonable computational effort. A case study model is set up in the Grasshopper environment and couples simulation components for outdoor thermal comfort and Life Cycle Assessment. It allows to identify multi-objective trade-offs for a high-dimensional space of urban configuration degrees of freedom such as outdoor vegetation, photovoltaics, and building characteristics. We compare the workflow results to other MOO algorithms and show how it can support decision making in urban planning at early design phases. In our case study, the tool chain was able to investigate the multi dimensional space of urban configurations. It systematically identified Pareto-optimal solutions therein and reduced the number of model evaluations significantly. The case study results showcase the trade-off between lifecycle-based global warming potential (GWP) and outdoor thermal comfort. We identified the number of trees and the coverage of the east and west façades with photovoltaics as the most sensitive parameters. The proposed process proves to be a powerful multi-criteria decision support tool for urban planners. It allows to identify and quantify the Pareto Front of competing urban target trade-offs at early design phases. Additionally, it visualizes them according to the boundary conditions of urban development. The input configurations of the obtained Pareto-solutions serve as a base of urban planning recommendations. In our case study, trees and photovoltaics prove to be good leverage points in the area of GWP optimal solutions. However, urban planners need to carefully coordinate inputs when aiming for a specific trade-off balance. The tool-chain and the simulation model offer further potential for investigating neighborhood typologies. Thereby, applicants can derive scalable guidance to support the sustainable transformation of the urban environment.

Systematizing heterogeneous expert knowledge, scenarios and goals via a goal-reasoning artificial intelligence agent for democratic urban land use planning

AI Agent as Urban Planner: Steering Stakeholder Dynamics in Urban Planning via Consensus-based Multi-Agent Reinforcement Learning

An intelligent spatial land use planning support system using socially rational agents

Spatial planning of urban communities via deep reinforcement learning

ReasonPlanner: Enhancing Autonomous Planning in Dynamic Environments with Temporal Knowledge Graphs and LLMs

Automated Urban Planning for Reimagining City Configuration via Adversarial Learning: Quantification, Generation, and Evaluation

Multi-Generative Agent Collective Decision-Making in Urban Planning: A Case Study for Kendall Square Renovation

A Generative Design Method for the Functional Layout of Town Planning Based on Multi-Agent System

Towards Automated Urban Planning: When Generative and ChatGPT-like AI Meets Urban Planning

Accelerating Cooperative Planning for Automated Vehicles with Learned Heuristics and Monte Carlo Tree Search

Urban-GAN: An artificial intelligence-aided computation system for plural urban design

A Human-Like Reasoning Framework for Multi-Phases Planning Task with Large Language Models

Empowering Scenario Planning with Artificial Intelligence: A Perspective on Building Smart and Resilient Cities

A Human-Centered Data-Driven Planner-Actor-Critic Architecture via Logic Programming

Urban systems exploration: A generic process for multi-objective urban planning to support decision making in early design phases

The Landscape of Emerging AI Agent Architectures for Reasoning, Planning, and Tool Calling: A Survey

RoboGPT: an intelligent agent of making embodied long-term decisions for daily instruction tasks

Agent Q: Advanced Reasoning and Learning for Autonomous AI Agents

Divide-and-Conquer Monte Carlo Tree Search For Goal-Directed Planning

LaMMA-P: Generalizable Multi-Agent Long-Horizon Task Allocation and Planning with LM-Driven PDDL Planner

Artificial intelligence for sustainable development of smart cities and urban land-use management