MapDreamer: Aerial Imagery Conditioned Latent Diffusion for Lane-Level Map Generation

• URL: http://arxiv.org/abs/2607.01370v1
• Date: Wed, 01 Jul 2026 18:33:38 GMT
• Title: MapDreamer: Aerial Imagery Conditioned Latent Diffusion for Lane-Level Map Generation
• Abstract summary: We present MapDreamer, a generative diffusion model that synthesizes lane-level vector maps with explicit topology directly from a single aerial image.<n>To align generated maps with the observed scene, we condition each denoising step on dense aerial features injected through cross-attention.<n>Experiments on UrbanLaneGraph derived from Argoverse 2 show improved geometric and topological fidelity over non-generative baselines.
• Score: 11.066720921275648
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: High definition map generation is essential for autonomous driving, yet remains a labor-intensive process at scale. We present MapDreamer, a generative diffusion model that synthesizes lane-level vector maps with explicit topology directly from a single aerial image. MapDreamer learns a compact latent representation of lane centerlines and their topological relations using a variational autoencoder and predicts graphs with a transformer-based latent diffusion model. To align generated maps with the observed scene, we condition each denoising step on dense aerial features injected through cross-attention. To handle the varying number of lanes across scenes, we propose a lane cardinality module paired with background ghost lane latents, a learned buffer that prevents slot collapse during diffusion. Furthermore, we introduce a sliding-window global graph aggregation strategy that stitches local tiles into city-scale maps while preserving connectivity through encoded lane boundaries. Experiments on UrbanLaneGraph derived from Argoverse 2 show improved geometric and topological fidelity over non-generative baselines.

Related papers

• Generative Lane Topology Reasoning via Autoregressive Model with Geometry Prior [35.786221225043455]
  TopoGPT is a generative framework that learns the geometry prior to typical lane graph structures.<n>We construct a large-scale map dataset comprising 3.3M scenes.<n>TopoGPT outperforms existing methods by an average of +6.4 on lane-level and +11.6 on point-level metrics.
  arXiv  Detail & Related papers  (2026-06-30T15:27:54Z)
• T2SG: Traffic Topology Scene Graph for Topology Reasoning in Autonomous Driving [26.038699227233227]
  Traffic Topology Scene Graph is a unified scene graph explicitly modeling the lane, controlled and guided by different road signals.<n>For the generation of T2SG, we propose TopoFormer, a novel one-stage Topology Scene Graph TransFormer with two newly designed layers.
  arXiv  Detail & Related papers  (2024-11-28T03:55:50Z)
• TopoSD: Topology-Enhanced Lane Segment Perception with SDMap Prior [70.84644266024571]
  We propose to train a perception model to "see" standard definition maps (SDMaps) We encode SDMap elements into neural spatial map representations and instance tokens, and then incorporate such complementary features as prior information. Based on the lane segment representation framework, the model simultaneously predicts lanes, centrelines and their topology.
  arXiv  Detail & Related papers  (2024-11-22T06:13:42Z)
• Diffusion-based Data Augmentation for Object Counting Problems [62.63346162144445]
  We develop a pipeline that utilizes a diffusion model to generate extensive training data. We are the first to generate images conditioned on a location dot map with a diffusion model. Our proposed counting loss for the diffusion model effectively minimizes the discrepancies between the location dot map and the crowd images generated.
  arXiv  Detail & Related papers  (2024-01-25T07:28:22Z)
• Augmenting Lane Perception and Topology Understanding with Standard Definition Navigation Maps [51.24861159115138]
  Standard Definition (SD) maps are more affordable and have worldwide coverage, offering a scalable alternative. We propose a novel framework to integrate SD maps into online map prediction and propose a Transformer-based encoder, SD Map Representations from transFormers. This enhancement consistently and significantly boosts (by up to 60%) lane detection and topology prediction on current state-of-the-art online map prediction methods.
  arXiv  Detail & Related papers  (2023-11-07T15:42:22Z)
• Path-Aware Graph Attention for HD Maps in Motion Prediction [4.531240717484252]
  Success of motion prediction for autonomous driving relies on integration of information from the HD maps. We propose Path-Aware Graph Attention, a novel attention architecture that infers the attention between two vertices by parsing the sequence of edges forming the paths that connect them. Our analysis illustrates how the proposed attention mechanism can facilitate learning in a didactic problem where existing graph networks like GCN struggle.
  arXiv  Detail & Related papers  (2022-02-23T09:43:47Z)
• HDMapGen: A Hierarchical Graph Generative Model of High Definition Maps [81.86923212296863]
  HD maps are maps with precise definitions of road lanes with rich semantics of the traffic rules. There are only a small amount of real-world road topologies and geometries, which significantly limits our ability to test out the self-driving stack. We propose HDMapGen, a hierarchical graph generation model capable of producing high-quality and diverse HD maps.
  arXiv  Detail & Related papers  (2021-06-28T17:59:30Z)
• DAGMapper: Learning to Map by Discovering Lane Topology [84.12949740822117]
  We focus on drawing the lane boundaries of complex highways with many lanes that contain topology changes due to forks and merges. We formulate the problem as inference in a directed acyclic graphical model (DAG), where the nodes of the graph encode geometric and topological properties of the local regions of the lane boundaries. We show the effectiveness of our approach on two major North American Highways in two different states and show high precision and recall as well as 89% correct topology.
  arXiv  Detail & Related papers  (2020-12-22T21:58:57Z)
• Learning Lane Graph Representations for Motion Forecasting [92.88572392790623]
  We construct a lane graph from raw map data to preserve the map structure. We exploit a fusion network consisting of four types of interactions, actor-to-lane, lane-to-lane, lane-to-actor and actor-to-actor. Our approach significantly outperforms the state-of-the-art on the large scale Argoverse motion forecasting benchmark.
  arXiv  Detail & Related papers  (2020-07-27T17:59:49Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.