{"slug": "mars-a-foundational-map-auto-regressor", "title": "MARS - A foundational map auto-regressor", "summary": "Researchers released MARS (Map Auto-Regressor), described as the first map foundation model, which generates multi-polyline road networks and polygon buildings in a unified end-to-end manner rather than through conventional multi-stage pixel segmentation and vectorized post-processing. Training used MAP-3M, a multi-class map extraction dataset of 3.4 million examples that the authors call the largest of its kind, and MARS outperformed or matched multistage baselines across four road and building datasets. The team also demonstrated a \"Chat with MARS\" capability for interactive human-in-the-loop map generation and correction, and released the MAP-3M dataset and a project demo page on Hugging Face.", "body_md": "arXiv:2610.20843v1 Announce Type: new \nAbstract: Map generation tasks feature extensive non-structural vectorized data (e.g., points, polylines, and polygons) and thus pose significant challenges to common pixel-wise generative models. Conventional approaches use multiple stages, first segmenting these features at the pixel level and then performing vectorized post-processing, with errors and complexity compounding at each stage. Motivated by the recent success of auto-regressive language modeling, we propose the first map foundation model, named Map Auto-Regressor (MARS), that is capable of generating both multi-polyline road networks and polygon buildings in a unified manner. For training MARS, we collected to our knowledge the largest multi-class map extraction dataset totaling 3.4M examples, which we call MAP-3M. Across four road and building datasets, MARS outperforms or matches the performance of multistage baselines. Additionally, we develop a \"Chat with MARS\" capability that enables interactive human-in-the-loop map generation and correction, supported by the auto-regressive nature of our end-to-end approach. We release our MAP-3M dataset and project demo page at (1) https://huggingface.co/datasets/bag-lab/MAP-3M and (2) https://huggingface.co/spaces/bag-lab/MARS, respectively.", "url": "https://wpnews.pro/news/mars-a-foundational-map-auto-regressor", "canonical_source": "https://arxiv.org/abs/2610.20843", "published_at": "2026-10-16 04:00:00+00:00", "updated_at": "2026-10-16 05:08:10.490176+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "ai-research", "generative-ai"], "entities": ["MARS", "Map Auto-Regressor", "MAP-3M", "Hugging Face", "bag-lab", "arXiv"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/mars-a-foundational-map-auto-regressor", "markdown": "https://wpnews.pro/news/mars-a-foundational-map-auto-regressor.md", "text": "https://wpnews.pro/news/mars-a-foundational-map-auto-regressor.txt", "jsonld": "https://wpnews.pro/news/mars-a-foundational-map-auto-regressor.jsonld"}}