SMART: Advancing Scalable Map Priors for Driving Topology Reasoning

2025-02-06

AI Summary

Existing lane topology inference methods for autonomous driving suffer from poor generalizability and limited scalability due to reliance on specific onboard sensor configurations. Method: This paper introduces the first fully sensor-agnostic offline topology inference framework, leveraging only standard-definition (SD) imagery and satellite maps. It employs multi-source map cross-scale alignment, geocoordinate-constrained self-supervised pretraining, HD-map distillation supervision, and a plug-and-play feature fusion architecture to learn robust map priors. Contribution/Results: The framework enables zero-shot integration with arbitrary online inference systems, significantly enhancing deployment flexibility and cross-platform generalization. On the OpenLane-V2 benchmark, it achieves a 28% improvement in topology recognition accuracy over state-of-the-art sensor-dependent methods using only SD imagery and satellite inputs—marking the first demonstration of high-accuracy, sensor-free lane topology inference in offline settings.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Search and Optimization: Learning to Search
• Computer Vision: Remote Sensing / Geospatial AI

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Topology reasoning is crucial for autonomous driving as it enables comprehensive understanding of connectivity and relationships between lanes and traffic elements. While recent approaches have shown success in perceiving driving topology using vehicle-mounted sensors, their scalability is hindered by the reliance on training data captured by consistent sensor configurations. We identify that the key factor in scalable lane perception and topology reasoning is the elimination of this sensor-dependent feature. To address this, we propose SMART, a scalable solution that leverages easily available standard-definition (SD) and satellite maps to learn a map prior model, supervised by large-scale geo-referenced high-definition (HD) maps independent of sensor settings. Attributed to scaled training, SMART alone achieves superior offline lane topology understanding using only SD and satellite inputs. Extensive experiments further demonstrate that SMART can be seamlessly integrated into any online topology reasoning methods, yielding significant improvements of up to 28% on the OpenLane-V2 benchmark.

Problem

• Eliminate sensor dependency in lane perception
• Enhance scalability of topology reasoning methods
• Improve autonomous driving with map prior models

Innovation

• Leverages SD and satellite maps
• Eliminates sensor-dependent features
• Enhances online topology reasoning methods