Online Temporal Fusion for Vectorized Map Construction in Mapless Autonomous Driving

TL;DR

This paper introduces an online system for vectorized map construction in autonomous driving that leverages long-term temporal data to improve map accuracy and robustness without relying on high-definition maps.

Contribution

It presents a novel method that fuses historical sensor data into a semantic voxel map and constructs an instance-level representation of road markings for better map consistency.

Findings

More accurate and consistent maps than existing methods

Effective in complex urban environments

Enhances closed-loop autonomous driving performance

Abstract

To reduce the reliance on high-definition (HD) maps, a growing trend in autonomous driving is leveraging onboard sensors to generate vectorized maps online. However, current methods are mostly constrained by processing only single-frame inputs, which hampers their robustness and effectiveness in complex scenarios. To overcome this problem, we propose an online map construction system that exploits the long-term temporal information to build a consistent vectorized map. First, the system efficiently fuses all historical road marking detections from an off-the-shelf network into a semantic voxel map, which is implemented using a hashing-based strategy to exploit the sparsity of road elements. Then reliable voxels are found by examining the fused information and incrementally clustered into an instance-level representation of road markings. Finally, the system incorporates domain knowledge to estimate the geometric and topological structures of roads, which can be directly consumed by the planning and control (PnC) module. Through experiments conducted in complicated urban environments, we have demonstrated that the output of our system is more consistent and accurate than the network output by a large margin and can be effectively used in a closed-loop autonomous driving system.