High-definition (HD) mapping tasks, which perform lane detections and predictions, are extremely challenging due to non-ideal conditions such as view occlusions, distant lane visibility, and adverse weather conditions. Those conditions often result in compromised lane detection accuracy and reduced reliability within autonomous driving systems. To address these challenges, we introduce MapATM, a novel deep neural network that effectively leverages historical actor trajectory information to improve lane detection accuracy, where actors refer to moving vehicles. By utilizing actor trajectories as structural priors for road geometry, MapATM achieves substantial performance enhancements, notably increasing AP by 4.6 for lane dividers and mAP by 2.6 on the challenging NuScenes dataset, representing relative improvements of 10.1% and 6.1%, respectively, compared to strong baseline methods. Extensive qualitative evaluations further demonstrate MapATM's capability to consistently maintain stable and robust map reconstruction across diverse and complex driving scenarios, underscoring its practical value for autonomous driving applications.
Figures & tables
Figure 1: Architecture of the proposed MapATM framework. The model combines multi view image features with actor trajectory information through dedicated modules for actor trajectory modeling and multi domain information fusion, resulting in more accurate online HD map construction.
Figure 2: Overall architecture of MapATM. Sensor data and transformed actor trajectories are first converted into BEV features. Learnable queries capture map elements (e.g., lanes) and actors (e.g., vehicles). Actor trajectories explicitly encode spatial-temporal map geometry, which is fused with map queries via cross-attention, resulting in enhanced map element inference.
Method
Dataset
APped
APdivider
APboundary
mAP
VAD
general
37.7
45.2
44.3
42.4
MapATM
general
41.1
49.8
44.1
45.0
VAD
complex
34.1
45.1
47.1
42.1
MapATM
complex
35.1
49.7
47.0
44.0
Table 1: Comparison on nuScenes val set. The APs of VAD are taken from the official checkpoints for all dataset, and fine tuned for complex dataset.
Figure 3: Comparison of Distance under different occlusion interval
Modeling method
APped
APdivider
APboundary
mAP
VAD
34.1
45.1
47.1
42.1
Pre-Fusion
33.1
45.7
43.8
40.9
Pre-Fusion + Projection
33.3
44.8
45.0
41.0
In-Fusion(Our)
35.1
49.7
47.0
44.0
Table 2: Ablations about multi-modality Fusion
Fusion method
APped
APdivider
APboundary
mAP
Layer-wise attention
30.5
48.4
46.8
41.9
DeformDETR [ deformDETR ]
35.1
49.7
47.0
44.0
Table 3: Ablations about Fusion Methods
Modeling method
APped
APdivider
APboundary
mAP
Backbone
33.5
47.8
47.0
42.7
Backbone + Encoder
32.4
47.6
45.9
41.9
Backbone + Actor query
35.1
49.7
47.0
44.0
Table 4: Ablations about modeling method of Actor Trajectory Modeling
Pretraining Output
APped
APdivider
APboundary
mAP
Virtual centerline
31.4
44.5
46.3
40.7
Virtual lane edge
35.1
49.7
47.0
44.0
Table 5: Ablations about actor quries of Actor Trajectory Modeling
Figure 4: Qualitative results under occlusion. The left vectorized HD map is from VAD, the right is MapATM, and the green and red boxes represent the ego car and surrounding cars respectively.
Accurate, up-to-date representations of road structures are critical for the safe operation of autonomous vehicles. Existing systems rely either on costly, maintenance-heavy high-definition (HD) maps which compromise safety when outdated, or purely sensor-based online mapping which struggles with long-range reliability and occlusion. Systems incorporating map prior information into online mapping seek to overcome drawbacks of both approaches by combining them in some way. We propose 'Driveline To HD Map' (D2HDMap), an online mapping system that injects a lightweight, non-visible driveline prior to guide the estimation of visible road structures such as lane dividers, road boundaries and crosswalks. This prior incurs less effort to create and update compared to full HD map priors used in other approaches. We also show that training with such a prior can improve generalization at inference time when no prior is available. Ablation studies conducted on the nuScenes and Argoverse 2 dataset demonstrate that models trained using a driveline prior largely retain performance even when priors are not available. On a geographically disjoint split, D2HDMap achieves 44.8 mAP, surpassing recent state-of-the-art. Additionally, noise-aware training substantially increases robustness to realistic localization error.
Seojun Shon, Chikao Tsuchiya, Dhaval Bhanderi +3
Nissan North America · Nissan Advanced Technology Center - Silicon Valley, Nissan North America, Santa Clara, CA 95014, USA
High-definition (HD) maps are essential for autonomous driving systems. In constructing such maps, onboard multi-view camera images, standard-definition maps and satellite images provide crucial information. However, due to the modality and perspective differences among these data sources, existing methods often struggle to effectively align and fuse them, making online HD map construction still challenging. To address these issues, we propose Driver2Map, an online HD map construction model inspired by human drivers. Unlike existing HD map construction models that utilize only two modalities, our Driver2Map can simultaneously exploit three modalities. Specifically, we propose a "two-stage alignment" strategy to reduce spatial misalignment across different modalities. Additionally, we introduce "Pose-Guided BEV Fusion", a BEV (bird's-eye-view) generation module that leverages camera pose information to adaptively weight multi-view features, thereby effectively suppressing cross-view feature overlap during BEV generation. Also, we design a "Pretrained Prior for Map Refinement" module to refine the initial prediction by learning map structure priors, thus improving the HD map prediction under dynamic occlusions. Extensive experiments demonstrate that Driver2Map outperforms existing methods on both IoU and AP metrics.
High-definition maps provide critical structural information for downstream autonomous driving tasks. Recent methods increasingly rely on external priors to improve prediction accuracy and robustness. Yet, existing prior representations expose a trade-off: dense rasters are easy to integrate but ignore certain map geometric properties, while sparse vectors preserve geometric structure but require dedicated encoders and are difficult to plug into existing architectures. Moreover, external priors can be inaccurate, costly to acquire, or difficult to maintain, motivating a structural representation that can provide geometric guidance while remaining learnable and broadly applicable. We address this limitation with the Lane Field, a dense BEV-aligned geometric representation that encodes local lane structure at each spatial location and can be either predicted from onboard observations or constructed from external map priors. Building on this representation, we introduce a plug-and-play framework that enhances online HD mapping through two complementary mechanisms: BEV feature modulation, which injects lane-structured cues into latent BEV features, and deformable attention guidance, which steers query sampling toward the underlying lane geometry. We primarily evaluate our method on OpenLane-V2, showing that conditioning BEV features and guiding deformable attention with local lane geometry enables our framework to consistently improve existing methods and achieve state-of-the-art performance on lane segment perception. Project page: https://andberto.github.io/LaneField/
Andrea Bertogalli, Anthony Dell'Eva, Marcello Ceresini +5
University of Parma, Italy · VisLab, Italy · University of Bologna, Italy