Integration of Visual SLAM into Consumer-Grade Automotive Localization

2025-11-10

AI Summary

To address systematic errors and calibration biases inherent in wheel odometry and IMUs for consumer-grade automotive localization, this paper proposes a novel online calibration framework that tightly integrates visual-inertial SLAM with a vehicle lateral dynamic model. Leveraging high-accuracy motion constraints from SLAM, the method enables real-time estimation and compensation of gyroscope bias—overcoming the limitations of conventional approaches relying solely on proprioceptive sensors. By fusing multi-source measurements (visual features, IMU readings, wheel speeds, and dynamic model constraints) in a tightly coupled manner, the framework achieves robust ego-motion estimation under complex driving conditions. Extensive experiments on both private and public benchmarks—including KITTI and Oxford RobotCar—demonstrate a 42% improvement in gyroscope zero-bias calibration accuracy and a 31% average reduction in absolute pose error compared to state-of-the-art methods, significantly enhancing both localization accuracy and long-term stability.

Technology Category

• Intelligent Robots: State Estimation
• Computer Vision: Motion & Tracking
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Accurate ego-motion estimation in consumer-grade vehicles currently relies on proprioceptive sensors, i.e. wheel odometry and IMUs, whose performance is limited by systematic errors and calibration. While visual-inertial SLAM has become a standard in robotics, its integration into automotive ego-motion estimation remains largely unexplored. This paper investigates how visual SLAM can be integrated into consumer-grade vehicle localization systems to improve performance. We propose a framework that fuses visual SLAM with a lateral vehicle dynamics model to achieve online gyroscope calibration under realistic driving conditions. Experimental results demonstrate that vision-based integration significantly improves gyroscope calibration accuracy and thus enhances overall localization performance, highlighting a promising path toward higher automotive localization accuracy. We provide results on both proprietary and public datasets, showing improved performance and superior localization accuracy on a public benchmark compared to state-of-the-art methods.

Problem

• Integrating visual SLAM into consumer vehicles to enhance localization accuracy
• Fusing visual SLAM with vehicle dynamics for online gyroscope calibration
• Overcoming limitations of wheel odometry and IMUs through vision-based integration

Innovation

• Fuses visual SLAM with lateral vehicle dynamics model
• Enables online gyroscope calibration during driving
• Improves localization accuracy using vision-based integration