Revision history [back]

I'd suggest looking into Hough Lines, as that's how Autonomous Vehicles currently detect lines.

In terms of setting the lines as an obstacle ... it depends on the use case. If you're using move_base to move a robot, then you would need obstacles yes. .. at any rate, you would need a camera to detect the lines. I'd suggest using a depth camera: use the color image to get your lines and then get the 3D points of those pixels by 'overlaying' them with the depth image. With those 3D points you can create obstacles. You'll likely need to look more into move_base to see how that could be done. I can see a shortcut where you just fake the obstacle by publishing to '/scan' so that move_base will see those points as lidar scans and thus avoid them.

Good luck