Sign recognition for autonomous vehicles

By | September 27, 2018

A method 300, performed by a vehicle controller, comprises: identifying clusters of points in a point cloud obtained 302 from a LIDAR (light detection and ranging) sensor having reflectivity above a reflectivity threshold 306; performing symbol and/or character recognition 314 for regions of an output of a camera corresponding 312 to one or more of the clusters of points; and refraining from performing symbol/character recognition for regions of the output of the camera not corresponding to the clusters of points. Clusters may further be selected according to a flatness threshold 310, and may be further refined by removing clusters corresponding to a road surface or ground plane 304. Autonomous vehicle navigation, including braking, accelerating, and steering, may be controlled by further performing obstacle detection using the point cloud. A vehicle is provided to perform the method, wherein a camera and LIDAR sensor are mounted to the vehicle. The invention reduces the field of view on which character recognition must be performed in order to reduce the computing burden.

Read the original article

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.