OCR in Robotics - Visual recognition of font based Landmarks for Mobile Systems

This video presents a novel approach, to solve the localization problem for mobile robots, by reading doorplates as non-artificial font based landmarks in indoor environments. This implementation utilizes a high resolution consumer camera, to detect and take pictures of the doorplate landmarks. The doorplates are found by a novel algorithm, which detects the layout of a doorplate in a viewfinder image (320x240) at 20 Hz in real-time. After the doorplate is found, the robot focuses the doorplate and takes a high resolution image with up to 9 mega pixels. Finally, the content of the doorplate is converted to text, by using standard OCR engines. This approach is the first important step towards reading robots, which act like humans and a revolution to standard robotic mapping.

Category:

Science & Technology

Tags:

• robotic
• ocr
• mobile robotic
• opencv
• semantic mapping
• image processing

[o0oD4n1so0o]

In my opinion u need better camera, and I would also like this thing working with a 20MHz processor of ATMega family.

I am also interested in the Algorithm if you could share it leave me a pm!