Simultaneous Localization and Mapping (SLAM) on NAO

Simultaneous Localization and Mapping (SLAM) is a navigation and mapping method used by autonomous robots and moving vehicles. SLAM is mainly concerned with the problem of building a map in an unknown environment and concurrently navigating through the environment using the map. Localization is of utmost importance to allow the robot to keep track of its position with respect to the environment and the common use of odometry proves to be unreliable. SLAM has been proposed as a solution by previous research to provide more accurate localization and mapping on robots. This project involves the implementation of the SLAM algorithm in the humanoid robot NAO by Aldebaran Robotics. The SLAM technique will be implemented using vision from the single camera attached to the robot to map and localize the position of NAO in the environment. The result details the attempt to implement specifically the chosen algorithm, 1-Point RANSAC Inverse Depth EKF Monocular SLAM by Dr Javier Civera on the robot NAO. The algorithm is shown to perform well for smooth motions but on the humanoid NAO, the sudden changes in motion produces undesirable results.This study on SLAM will be useful as this technique can be widely used to allow mobile robots to map and navigate in areas which are deemed unsafe for humans.