# MapTrace: Scalable Data Generation for Route Tracing on Maps

arXiv: https://arxiv.org/abs/2512.19609 · HTML: https://mlsanity.com/paper/2512.19609/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Embodied and 3D vision-language models, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 16 of its 35 references that are in the map.
- First posted on arXiv 22 Dec 2025.

## Placed in

- Multimodal LLMs / vision-language models / [Embodied and 3D vision-language models](https://mlsanity.com/mllm/embodied-and-3d-vision-language-models/)

## Builds on (most cited first, 16 of 16)

- [GLM-4.5V and GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning](https://mlsanity.com/paper/2507.01006/) (2025)
- [What's "up" with vision-language models? Investigating their struggle with spatial reasoning](https://mlsanity.com/paper/2310.19785/) (2023)
- [On LLMs-Driven Synthetic Data Generation, Curation, and Evaluation: A Survey](https://mlsanity.com/paper/2406.15126/) (2024)
- [ALLaVA: Harnessing GPT4V-Synthesized Data for Lite Vision-Language Models](https://mlsanity.com/paper/2402.11684/) (2024)
- [General Evaluation for Instruction Conditioned Navigation using Dynamic Time Warping](https://mlsanity.com/paper/1907.05446/) (2019)
- [Evaluating Spatial Understanding of Large Language Models](https://mlsanity.com/paper/2310.14540/) (2023)
- [CubiCasa5K: A Dataset and an Improved Multi-Task Model for Floorplan Image Analysis](https://mlsanity.com/paper/1904.01920/) (2019)
- [DetToolChain: A New Prompting Paradigm to Unleash Detection Ability of MLLM](https://mlsanity.com/paper/2403.12488/) (2024)
- [Self-Boosting Large Language Models with Synthetic Preference Data](https://mlsanity.com/paper/2410.06961/) (2024)
- [IsoBench: Benchmarking Multimodal Foundation Models on Isomorphic Representations](https://mlsanity.com/paper/2404.01266/) (2024)
- [Can Large Vision Language Models Read Maps Like a Human?](https://mlsanity.com/paper/2503.14607/) (2025)
- [Visual Position Prompt for MLLM based Visual Grounding](https://mlsanity.com/paper/2503.15426/) (2025)
- [Semantic Map-based Generation of Navigation Instructions](https://mlsanity.com/paper/2403.19603/) (2024)
- [Vision Language Models Can Parse Floor Plan Maps](https://mlsanity.com/paper/2409.12842/) (2024)
- [LLM-Guided Indoor Navigation with Multimodal Map Understanding](https://mlsanity.com/paper/2503.11702/) (2025)
- [SCRAMBLe : Enhancing Multimodal LLM Compositionality with Synthetic Preference Data](https://mlsanity.com/paper/2504.04740/) (2025)

## Built on by (most cited first, 3 of 3)

- [ICDAR2026 Competition on Multimodal Reasoning over Documents in Multiple Domains](https://mlsanity.com/paper/2609.25055/) (2026)
- [TransitLM: A Large-Scale Dataset and Benchmark for Map-Free Transit Route Generation](https://mlsanity.com/paper/2605.22355/) (2026)
- [Perceive, Interact, Reason: Building Tool-Augmented Visual Agents for Spatial Reasoning](https://mlsanity.com/paper/2606.12830/) (2026)

## Abstract (arXiv)

While Multimodal Large Language Models have achieved human-like performance on many visual and textual reasoning tasks, their proficiency in fine-grained spatial understanding, such as route tracing on maps remains limited. Unlike humans, who can quickly learn to parse and navigate maps, current models often fail to respect fundamental path constraints, in part due to the prohibitive cost and difficulty of collecting large-scale, pixel-accurate path annotations. To address this, we introduce a scalable synthetic data generation pipeline that leverages synthetic map images and pixel-level parsing to automatically produce precise annotations for this challenging task. Using this pipeline, we construct a fine-tuning dataset of 23k path samples across 4k maps, enabling models to acquire more human-like spatial capabilities. Using this dataset, we fine-tune both open-source and proprietary MLLMs. Results on MapBench show that finetuning substantially improves robustness, raising success rates by up to 6.4 points, while also reducing path-tracing error (NDTW). These gains highlight that fine-grained spatial reasoning, absent in pretrained models, can be explicitly taught with synthetic supervision.