Simultaneous Localization and Mapping for Autonomous Robot Navigation

Simultaneous localization and mapping (SLAM) is a challenging vision-based task that provides the ability for an autonomous robot to simultaneously map the environment and localized itself concerning it. It is a joint estimation task and cannot be decoupled. Robot operating system (ROS) provides many algorithmic and communicating nodes to work with autonomous robots. Several algorithms have been developed to solve this intractable time problem in unknown real-time environments assuming different constraints. In this paper, we introduce a new computer vision-based technique for SLAM using an extended Kalman filter for efficiently generating the map of the environment and navigate in the environment autonomously. The proposed algorithm in this paper effectively generates the shortest collision-free path to read the goal. The sensor used in this project is the Kinect sensor which is a cost-effective sensor having a camera and a depth sensor that uses infrared rays. The performance of the autonomous navigation is also presented in terms of the path following accuracy and obstacle avoidance.