From Static to Dynamic: a Survey of Topology-Aware Perception in Autonomous Driving

2025-09-28

AI Summary

Traditional static high-definition maps suffer from high construction costs, infrequent updates, and poor generalization—severely limiting the scalability of autonomous driving systems. To address these limitations, this paper proposes a topology-aware dynamic environment understanding framework. The framework integrates vectorized map construction, topological graph neural network modeling, multimodal large language model–guided semantic reasoning, and real-time sensor fusion to achieve compact, interpretable representations of lane geometry and road semantics, along with relational reasoning over structural and semantic elements. Compared to static-map paradigms, our approach significantly improves perception latency (millisecond-level map updates), cross-region generalization, and online adaptability. Experimental results demonstrate superior performance over state-of-the-art methods in dynamic scene modeling accuracy, map lightweighting (compression ratio >60%), and semantic consistency. This work establishes a novel paradigm for scalable, adaptive, and interpretable autonomous driving perception, supported by both theoretical foundations and empirical validation.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

The key to achieving autonomous driving lies in topology-aware perception, the structured understanding of the driving environment with an emphasis on lane topology and road semantics. This survey systematically reviews four core research directions under this theme: vectorized map construction, topological structure modeling, prior knowledge fusion, and language model-based perception. Across these directions, we observe a unifying trend: a paradigm shift from static, pre-built maps to dynamic, sensor-driven perception. Specifically, traditional static maps have provided semantic context for autonomous systems. However, they are costly to construct, difficult to update in real time, and lack generalization across regions, limiting their scalability. In contrast, dynamic representations leverage on-board sensor data for real-time map construction and topology reasoning. Each of the four research directions contributes to this shift through compact spatial modeling, semantic relational reasoning, robust domain knowledge integration, and multimodal scene understanding powered by pre-trained language models. Together, they pave the way for more adaptive, scalable, and explainable autonomous driving systems.

Problem

• Shifting from static maps to dynamic sensor-driven perception
• Overcoming costly static map limitations through real-time topology reasoning
• Enhancing autonomous driving with adaptive and scalable environmental understanding

Innovation

• Dynamic sensor-driven perception replaces static maps
• Compact spatial modeling with semantic relational reasoning
• Multimodal scene understanding using pre-trained language models