A Real-Time Lane Extraction System for the Wheelchair Robot

This paper presents a robust and reliable lane extraction algorithm to navigate the wheelchair robot in real time in outdoor environments. Seven components have been cooperatively utilized to extract lane boundaries on roads with different conditions. Specifically, three features are efficiently extracted to represent two lane boundaries using the lane boundaries in the previous frame. Three weights related to the three features are computed offline by the genetic algorithm for more accurate lane extraction. Furthermore, the average road intensity in all previous frames is also used to aid the robot in steering to the right direction if the inference-based verification system indicates consistent discrepancies between the current and previous frames. The system has been extensively tested and experimental results demonstrate our system can accurately guide the wheelchair robot in the navigating tasks in real time.