LMC-VIO: Lane Model-Constrained Monocular Inertial Visual SLAM for High-Precision Localization in Highway Scenes
Man Luo, Maosheng Yan, Yuan Guo, Bijun Li, Jian Zhou
AI summary
Problem
High-speed highway driving causes image blur and feature degradation, leading to scale estimation biases and heading drift in traditional visual-inertial SLAM systems. Current solutions often rely on expensive equipment or lack robustness in repetitive road scenarios.
Approach
LMC-VIO detects and tracks lane lines in camera images, then uses prior standard-definition map data to constrain camera scale, roll, and yaw through inverse perspective mapping and point-to-map matching before fusing these constraints into a monocular visual-inertial odometry framework.
Key results
- Novel lane model-constrained parameter calibration for accurate scale and roll estimation
- Point-to-map matching module that constrains vehicle heading using prior map data
- Seamless integration of lane constraints into VINS-Fusion for real-time localization optimization
- Demonstrated significant localization accuracy improvements on a self-collected highway dataset
Why it matters
Enables standard vehicles with basic cameras and IMUs to achieve high-precision highway localization without costly LiDAR or high-definition maps, advancing cost-effective autonomous driving deployment.
Abstract
Continuous stability, as one of the core modules of the autopilot system, is particularly important for its per- formance. However, as the vehicle speed increases, the system positioning error may be amplified, consequently introducing deviations in the positioning consistency of the system. The inherent high speeds and motion constraints in highway envi- ronments introduce new challenges for feature matching, par- ticularly in vision-based vehicle localization, where initialization and scale estimation biases are further expanded. Lane mark- ings, characterized by their simple and uniform structures and high distinctiveness from the surrounding environment, serve as effective features for matching-based localization in autonomous driving. This paper introduces a high-precision and robust vehicle localization method based on lane model constraints. Initially, leveraging lane model parameters from prior maps, we track and model lane line detections across consecutive frames to enhance the completeness of lane representation. The tracking results, combined with prior map data on lane widths, are employed to optimize scale parameters. Subsequently, real-time detected lanes are matched with prior maps through point-map association to constrain the vehicle’s heading angle. Finally, map matching results are integrated into existing visual local odometry methods to perform real-time localization optimization, thereby improving localization performance. Experimental evaluations conducted on a self-collected highway dataset demonstrate that the incorporation of lane models significantly enhances system localization accuracy.