Delay-Compensated Lane-Coordinate Vehicle State Estimation Using Low-Cost Sensors

Area of Science:

• Robotics and Control Systems
• Computer Vision for Autonomous Systems
• Sensor Fusion for Vehicle Dynamics

Background:

• Robust vehicle state estimation in a lane coordinate system is vital for Advanced Driver Assistance Systems (ADAS) and autonomous driving.
• Low-cost sensors like cameras, IMUs, and steering angle sensors present challenges due to complex vehicle dynamics and vision system signal delays.
• Accurate real-time estimation of lateral position, velocity, and heading angle is difficult to achieve with existing methods.

Purpose of the Study:

• To develop a lane-coordinate-based vehicle state estimator that overcomes the limitations of low-cost sensors and signal delays.
• To enable reliable and continuous estimation of critical vehicle states for ADAS and autonomous driving applications.
• To improve the accuracy and reduce the phase lag in vehicle state estimation compared to conventional methods.

Main Methods:

• Integration of a vehicle dynamics-based bicycle model with an Extended Kalman Filter (EKF).
• Implementation of a signal delay compensation algorithm to predict vehicle state evolution during camera processing.
• Development of an efficient camera processing pipeline with pre-trained networks for lane segmentation and state extraction.

Main Results:

• The proposed estimator provides continuous, high-accuracy, and delay-compensated lane-coordinate-based vehicle states.
• Significant reduction in estimation errors and phase lag compared to camera-only methods and estimators without delay compensation.
• Successful validation using real vehicle driving data on both straight and curved roads.

Conclusions:

• The developed estimator effectively addresses challenges in lane-based vehicle state estimation using low-cost sensors.
• The method enables reliable, real-time acquisition of critical vehicle state information for ADAS and autonomous driving.
• The approach offers a practical solution for enhancing the safety and performance of automated driving systems.