Robust online map building improves self-driving car navigation
PriorMaskMap: Robust Online Vectorized Map Construction with Biased Priors
Computer Vision and Pattern RecognitionRobotics
Summary
Mapping the environment in real time is critical for self-driving cars to understand the road and surroundings. Previous systems that use older maps to help can struggle when those maps are outdated or wrong. The authors introduce PriorMaskMap, which learns to judge the trustworthiness of these prior maps and selectively uses them to create more accurate, up-to-date maps. Their method is more precise and reliable, especially when prior maps contain errors.
What this means in practice
- •For autonomous vehicle developers: Build more accurate and reliable real-time maps for self-driving cars that can handle outdated or incorrect prior map data.
- •For robotics navigation teams: Improve robustness of robotic mapping systems that rely on integrating old maps with current visual inputs in changing environments.
Authors
Haoming Xu, Wei Li, Yu Hu
Abstract
Online vectorized map construction provides autonomous vehicles with essential, real-time semantic and geometric scene understanding. While leveraging prior maps presents an opportunity to improve accuracy, they are often biased due to outdated information or temporary road changes. Existing methods, which typically treat priors as fully reliable or only consider limited noise patterns, suffer severe performance degradation when these priors are imperfect. To address this, we propose PriorMaskMap, a robust framework for prior-guided vectorized map construction. Its first innovation is a Prior-Aware Confidence Estimator (PACE) that evaluates the input rasterized map against current visual observations. It assigns confidence scores to regions classified as reliable, erroneous, or blank, creating a foundational mask that guides two subsequent branches. In one branch, the Dual-Prior Query Generator (DPQG) selectively filters and embeds both rasterized and vectorized priors to initialize informative queries from trustworthy map structures. In the other, the Confidence-Guided Prior Fusion (CGPF) module spatially propagates and refines the mask, using the refined version to govern the fusion of BEV features with the prior map, thereby providing rich contextual representations for the map decoder. Evaluation on the nuScenes dataset shows that PriorMaskMap achieves superior mapping precision over existing approaches and exhibits marked robustness when provided with priors containing apparent errors. Our code is released at https://github.com/healenrens/PriorMaskMap.