3D cognitive modelling with mobile robots

This video shows a small mobile robot (Roomba), a 2D laserscanner (lms 100) and a standard digital camera (Canon). These components are combined to build 3D cognitive models. At first, the robots laserscanner takes a 3D scan with about 300.000 points.Then the point cloud is segmented into different regions, e.g walls, floors and ceiling. Depending on the laserscanners remission values, possible doorplate candidates ae calculated. The robot drives to the doorplate candidate to make high resolution pictures. Then OCR use those pictures to get the textual information out of those doorplates. This information is then merged with the 3D point cloud, to display the doorplate information. The results of the study were that, the classical simultaneous localization and mapping (Slam), and the kidnapped robot problem is easily solved with this human like approach. Finally this is an important step to intelligent robotics, and a small revolution, since the approach is simple, universal and not expensive, because it only uses commercial hardware.

Category:

Science & Technology

Tags:

• SLAM
• mobile robots
• OCR
• 3D laser scanner
• computer vision
• reading
• robots