NAO Robot Detects Objects Using Camera Images

Researchers at the Humanoid Robots Lab of Albert-Ludwigs-Universität Freiburg have made modifications to the NAO robot. The resulting NAO robot was showcased at ICRA 2011 conference.

This laboratory designed the laser range finder attachment in collaboration with Aldebaran Robotics. The data from the laser range finder attachment was combined with the monocular vision capabilities of the NAO robot to detect obstacles.

The laser range finder is placed on the top of NAO’s head. It is therefore not possible to spot obstacles present just in front of the robot. As a result, the robot has to stop from time to time and scan to locate the objects. The new approach developed by researchers enables the NAO robot to save time by depending on only camera images and not on laser scans. The robot starts scanning the ground and anything present above the ground level will be considered as an obstacle. It will then categorize objects based on their texture and color. The NAO robot can detect the objects with the help of images captured by the camera. It plans its path by building an occupancy grid.

Source: https://ras.papercept.net/