SenseWalk Agent-Based Semantic Trajectory Simulation Powered by Large Language Models in Zoned Environments

Jul 16, 2026

Semantic trajectory analysis models human movement by leveraging semantic information such as visitor profiles and goals, going beyond raw spatial paths to explain why people move in particular ways. Real-world semantic-trajectory analysis remains hindered by costly data collection and insufficient semantic richness, while existing simulation tools demand substantial technical expertise. SenseWalk introduces an interactive, LLM-powered agent system that simulates semantic trajectories in zoned environments using an LLM-and-social-force workflow, supported by a configurable interface, quantitative evaluation, and a user study.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages18

Reading Time45 min

Introduction

Semantic trajectories and their annotation types

Challenges in collecting semantic trajectory data

Motivation and goals for SenseWalk

SenseWalk: Agent-Based Semantic Trajectory Simulation Powered by Large Language Models in Zoned Environments Ziyue Lin [ziyuelin917@gmail.com](ziyuelin917@gmail.com)[ ](ziyuelin917@gmail.com)School of Data Science, Fudan University Shanghai, China Xinhang Xie School of Data Science, Fudan University Shanghai, China arXiv :2607 .00989v 1 [ cs .HC] 1 Jul 2026 Kangyi Wang School of Data Science, Fudan University Shanghai, China Siming Chen∗ [simingchen@fudan.edu.cn](simingchen@fudan.edu.cn)[ ](simingchen@fudan.edu.cn)School of Data Science, Fudan University Shanghai, China Abstract Semantic trajectory analysis has recently emerged as an approach for modeling human movement by capturing implicit patterns and behaviors through semantic information (e.g., visitors’ profiles and goals) beyond raw spatial paths to better understand why people move in certain ways. However, analyzing semantic trajectories in realworld scenarios remains challenging, as collecting high-quality data is costly and often lacks rich semantic information. Meanwhile, existing simulation tools require substantial technical expertise, which makes them difficult for practitioners to adopt. To address these limitations, the paper proposes 􀀨􀀴􀀽􀁂􀀴􀀬 􀀰􀀻􀀺, an interactive system that supports simulating semantic trajectories by LLM-powered agents. We develop a simulation workflow that combines LLMs and the social force model to balance physical plausibility and semantic coherence. A user-friendly interface is designed to facilitate users in customizing the simulation configuration and analyzing simulation outputs. We also conduct a quantitative experiment to evaluate the effectiveness of our simulation workflow, and a user study (n=12) to assess the usefulness and efficiency of our system. CCS Concepts • Human-centered computing → Interactive systems and tools. Keywords semantic trajectory, LLM-powered agent, interactive system ACM Reference Format: Ziyue Lin, Xinhang Xie, Kangyi Wang, and Siming Chen. 2018. SenseWalk: Agent-Based Semantic Trajectory Simulation Powered by Large Language Models in Zoned Environments. In Proceedings of Make sure to enter the correct conference title from your rights confirmation email ∗ Siming Chen is the corresponding author. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission [and/or a fee. Request permissions from permissions@acm.org](and/or a fee. Request permissions from permissions@acm.org). Conference acronym ’XX, Woodstock, NY © 2018 Copyright held by the owner/author(s) . Publication rights licensed to ACM. ACM ISBN 978-1-4503-XXXX-X/2018/06 [https://doi.org/XXXXXXX.XXXXXXX](https://doi.org/XXXXXXX.XXXXXXX) (Conference acronym ’XX). ACM, New York, NY, USA, 18 pages. [https:](https:)//[doi.org/XXXXXXX.XXXXXXX](doi.org/XXXXXXX.XXXXXXX) 1 Introduction Imagine a museum professional seeking to understand how different types of visitors navigate and engage with exhibitions. For instance, a first-time tourist may follow on-site signage and focus primarily on iconic artworks, whereas a local art student who frequently visits might adopt a self-directed route, selectively revisiting favored sections. Analyzing these distinct patterns holds substantial promise for personalized route recommendation [3, 11], adaptive exhibition design [65], and nuanced engagement evaluation [42], which are increasingly important for cultural institutions and public venues [3, 17, 23, 64] . Focusing solely on raw spatial paths or aggregate footfa

FAQs

What problem does semantic trajectory analysis aim to address?

It aims to model human movement using semantic signals like behaviors and goals, enabling explanations of movement patterns rather than relying only on spatial paths or aggregated counts.

Why is semantic trajectory analysis difficult in real-world scenarios?

High-quality spatiotemporal tracking is costly and noisy, semantics are not directly observable and need post-hoc annotation, visitor profiles are hard to obtain at scale due to privacy concerns, and cold-start settings can lack historical data.

How does SenseWalk simulate semantic trajectories for practical use?

SenseWalk provides an interactive system that uses LLM-powered agents and a workflow combining LLMs with the social force model to balance physical plausibility and semantic coherence, along with an interface for configuration and output analysis.

Jul 16, 2026

Upload by Olivia Brown