AgentSense: LLMs Empower Generalizable and Explainable Web-Based Participatory Urban Sensing
This work addresses the problem of deploying adaptive and explainable urban sensing systems for urban management, representing a novel method for a known bottleneck.
The paper tackles the limited generalization and poor interpretability of web-based participatory urban sensing systems by introducing AgentSense, a hybrid, training-free framework that integrates LLMs through a multi-agent evolution system, achieving distinct advantages in adaptivity and explainability over traditional methods and outperforming single-agent LLM baselines in performance and robustness.
Web-based participatory urban sensing has emerged as a vital approach for modern urban management by leveraging mobile individuals as distributed sensors. However, existing urban sensing systems struggle with limited generalization across diverse urban scenarios and poor interpretability in decision-making. In this work, we introduce AgentSense, a hybrid, training-free framework that integrates large language models (LLMs) into participatory urban sensing through a multi-agent evolution system. AgentSense initially employs classical planner to generate baseline solutions and then iteratively refines them to adapt sensing task assignments to dynamic urban conditions and heterogeneous worker preferences, while producing natural language explanations that enhance transparency and trust. Extensive experiments across two large-scale mobility datasets and seven types of dynamic disturbances demonstrate that AgentSense offers distinct advantages in adaptivity and explainability over traditional methods. Furthermore, compared to single-agent LLM baselines, our approach outperforms in both performance and robustness, while delivering more reasonable and transparent explanations. These results position AgentSense as a significant advancement towards deploying adaptive and explainable urban sensing systems on the web.