A lane-level localization method based on fusion of multiple ground signs

Abstract

The invention discloses a lane-level positioning method based on the fusion of multiple ground signs, comprising: step 1, acquiring vehicle road images; step 2, performing grayscale processing and filtering processing on the road images; step 3, processing in step 2 Carry out lane line detection on the road image after; Step 4, carry out stop line detection to road image; Step 5, carry out zebra crossing detection to road image; Step 6, locate the position of vehicle in the lane in real time according to lane line, zebra crossing and stop line. The present invention fuses the information of multiple marking lines on the road surface to accurately locate road lanes. It is not only applicable to the driving safety warning function of vehicles, but also can be integrated into the pure vision system in unmanned driving to perform lane inspection and turn at intersections.

Description

technical field

[0001] The invention belongs to the field of intelligent assisted driving technology and automatic driving technology, and in particular relates to a lane-level positioning method based on fusion of multiple ground signs. Background technique

[0002] In recent years, advancements in autonomous driving technology have attracted the attention and interest of researchers, the media, and the general public. In 2015, Google's self-driving car officially drove on the roads of California, USA for testing. Other major automakers and companies have generally acknowledged that it is only a matter of time before self-driving cars enter people's lives and profoundly affect people's living conditions. Vehicle localization plays a fundamental and critical role in intelligent transportation systems as it is a higher level pre-mission operation. Previously, most positioning methods relied on GNSS and inertial navigation systems, but when GPS loses signal under certain cir...

Images