Intelli-Planner: Towards Customized Urban Planning via Large Language Model Empowered Reinforcement Learning

TL;DR

This work addresses the need for participatory, customized urban planning by integrating large language models with deep reinforcement learning. It introduces Intelli-Planner, an Actor-Critic framework where LLMs formulate planning objectives and provide domain knowledge to the policy, while a multi-criteria evaluator guides learning through a five-dimensional reward system (service, ecology, economy, equity, satisfaction). The approach combines a knowledge-enhanced policy with an LLM-based stakeholder scoring mechanism and evaluates schemes across three diverse cities, showing improved objective metrics and higher stakeholder satisfaction versus baselines and existing methods. The results illustrate the practical potential of fusing LLM capabilities with DRL to streamline urban functional-area planning and better reflect stakeholder needs.

Abstract

Effective urban planning is crucial for enhancing residents' quality of life and ensuring societal stability, playing a pivotal role in the sustainable development of cities. Current planning methods heavily rely on human experts, which are time-consuming and labor-intensive, or utilize deep learning algorithms, often limiting stakeholder involvement. To bridge these gaps, we propose Intelli-Planner, a novel framework integrating Deep Reinforcement Learning (DRL) with large language models (LLMs) to facilitate participatory and customized planning scheme generation. Intelli-Planner utilizes demographic, geographic data, and planning preferences to determine high-level planning requirements and demands for each functional type. During training, a knowledge enhancement module is employed to enhance the decision-making capability of the policy network. Additionally, we establish a multi-dimensional evaluation system and employ LLM-based stakeholders for satisfaction scoring. Experimental validation across diverse urban settings shows that Intelli-Planner surpasses traditional baselines and achieves comparable performance to state-of-the-art DRL-based methods in objective metrics, while enhancing stakeholder satisfaction and convergence speed. These findings underscore the effectiveness and superiority of our framework, highlighting the potential for integrating the latest advancements in LLMs with DRL approaches to revolutionize tasks related to functional areas planning.

Figures (8)

• Figure 1: The overall architecture of Intelli-Planner. It follows an Actor-Critic architecture, including an Actor network that makes decisions on functional area types, and a Critic network that is continuously optimized through evaluation system.
• Figure 2: Formulate planning objectives according to the target style and the area's fundamental information.
• Figure 3: Knowledge enhancement of the decisions.
• Figure 4: Stakeholders are providing satisfaction scores based on planning demands and current conditions.
• Figure 5: The original functional types distribution in the planning regions of Beijing, Chicago, and Madrid.

Theorems & Definitions (1)

• Definition 1