# TrajMind: Chaining Role-Specialized LoRAs for Fast-and-Slow Collective Trajectory Anomaly Diagnosis

arXiv: https://arxiv.org/abs/2609.02540 · HTML: https://mlsanity.com/paper/2609.02540/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 15 of its 44 references that are in the map.
- First posted on arXiv 2 Sep 2026.

## Builds on (most cited first, 15 of 15)

- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [LoraHub: Efficient Cross-Task Generalization via Dynamic LoRA Composition](https://mlsanity.com/paper/2307.13269/) (2023)
- [Self-supervised Trajectory Representation Learning with Temporal Regularities and Travel Semantics](https://mlsanity.com/paper/2211.09510/) (2022)
- [VideoMind: A Chain-of-LoRA Agent for Temporal-Grounded Video Reasoning](https://mlsanity.com/paper/2503.13444/) (2025)
- [Disentangled Contrastive Learning for Social Recommendation](https://mlsanity.com/paper/2208.08723/) (2022)
- [UniTraj: Learning a Universal Trajectory Foundation Model from Billion-Scale Worldwide Traces](https://mlsanity.com/paper/2411.03859/) (2024)
- [Policy and World Modeling Co-Training for Language Agents](https://mlsanity.com/paper/2606.02388/) (2026)
- [Learning Representation for Anomaly Detection of Vehicle Trajectories](https://mlsanity.com/paper/2303.05000/) (2023)
- [Trajectory Anomaly Detection with Language Models](https://mlsanity.com/paper/2409.15366/) (2024)
- [CoBAD: Modeling Collective Behaviors for Human Mobility Anomaly Detection](https://mlsanity.com/paper/2508.14088/) (2025)
- [Dataset Condensation for Recommendation](https://mlsanity.com/paper/2310.01038/) (2023)
- [TransferTraj: A Vehicle Trajectory Learning Model for Region and Task Transferability](https://mlsanity.com/paper/2505.12672/) (2025)
- [Seeing Farther and Smarter: Value-Guided Multi-Path Reflection for VLM Policy Optimization](https://mlsanity.com/paper/2602.19372/) (2026)
- [Traj-MLLM: Can Multimodal Large Language Models Reform Trajectory Data Mining?](https://mlsanity.com/paper/2509.00053/) (2025)
- [GADformer: A Transparent Transformer Model for Group Anomaly Detection on Trajectories](https://mlsanity.com/paper/2303.09841/) (2023)

## Abstract (arXiv)

Diagnosing collective anomalies from urban trajectories is increasingly important for traffic governance, as it reveals what happened, who was involved, and where and when the event occurred. Existing detectors efficiently produce scores or labels, whereas vision--language pipelines provide richer semantics; neither couples verifiable diagnosis with low-latency monitoring. The central challenge is to recognize collective patterns and recover exact event details from the source trajectories without running the full diagnostic pipeline for every monitored window. We therefore separate always-on screening from on-demand diagnosis: screening raises alerts, while diagnosis releases only source-verified what--who--where--when records. We present TrajMind, a fast-and-slow framework that switches three role-specialized LoRA adapters over one frozen vision--language backbone. Its slow path, textitTrajMindslow, chains canvas-based typing, type-conditioned localization over serialized trajectories, and executable verification, yielding structured, evidence-backed diagnoses. Additionally, the fast path, textitTrajMindfast, screens each window in a single text-only pass, delivering efficient structured alerts. Extensive experiments show that, TrajMindslow outperforms the strongest baselines by at least 15.3 percentage points in anomaly typing and 13.8 percentage points in localization. These gains persist under cross-city transfer, and TrajMindfast reduces latency by 41.1% and maintains binary balanced accuracy of at least 93.5%. Together, TrajMind delivers accurate, evidence-backed diagnoses across cities and efficient front-line monitoring.