Abstract
For automated driving technology to move beyond the proof stage and into actual automated driving, it is important to verify the safety of the automated driving system. In this paper, a method to ensure lateral control safety for the lane detection function is proposed to address a failure of the image-sensor-based automated driving system, which is the system with the highest possibility for practical mass production. The proposed algorithm consists of three parts: the first part is the ego-motion estimator that estimates the movement of a vehicle; the second is an integrated lane detection sensor module, which response to failure by estimating lane information at the corresponding time; the last is a lane estimator, which tracks lane coefficients based on constant lane widths. A combination of these three modules and the prediction of the lane coefficient, C3, in the virtual sensor, which was not reflected in our previous study, enables a more robust response to lane detection failure. The performance difference between the proposed algorithm and an existing algorithm was confirmed by simulated evaluations for identical situations based on actual data. Through this result, it was confirmed that not only could near-lane estimation accuracy but also lane estimation accuracy at a far distance could be improved when compared to the existing algorithm. The results of this study are expected to help obtain lateral safety during minimal risk maneuver. The proposed algorithm may also contribute to ensuring the safety of image-sensor-based automated driving systems.
Highlights
Automated driving system (ADS) technology has advanced rapidly over the past decade owing to advances in other fundamental technologies, such as sensors and computing modules for self-driving, and the efforts of numerous researchers
Lane estimation accuracy is improved by estimating curvature-rate, which was not estimated in the existing virtual sensor, using vehicle dynamic information
According to the aspect of the result values, as the vehicle longitudinal/lateral velocity, lateral acceleration, and Z-axis angular velocity that were obtained through the ego-motion estimation are reflected, the lane coefficients, C0 ∼ C2, which were derived through the proposed algorithm, exhibit a similar detection pattern as the sensor values
Summary
Automated driving system (ADS) technology has advanced rapidly over the past decade owing to advances in other fundamental technologies, such as sensors and computing modules for self-driving, and the efforts of numerous researchers. In the vision-sensor-based path following method, the most important function is to detect the lane that the vehicle will follow. A lane tracking algorithm based on a stereo vision sensor is proposed to estimate a lane more robustly in the event of lane detection failure. The basic principle of the virtual sensor is to estimate the position of the current lane by inversely calculating the movement of the vehicle based on the last reliable detection value. For this reason, it is important to estimate the exact ego-motion of the vehicle. Lane estimation accuracy is improved by estimating curvature-rate, which was not estimated in the existing virtual sensor, using vehicle dynamic information
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
