Multi-Sensor SLAM for efficient Navigation of a Mobile Robot

Abstract

In this work, localization of the landmarks has been solved without prior knowledge of the environment. A well-known SLAM technique (Extended Kalman Filter and RGBD-SLAM) has been used to solve the localization of landmarks and to build 2D and 3D maps of the environment. SLAM techniques are implemented on a two-wheeled mobile robot by using an encoder to measure the feedback. The robot is programmed intelligently to autonomously navigate in an indoor static environment. A Sonar sensor is installed for for obstacle avoidance which reduces the computational cost. LiDAR and Microsoft Kinect (RGBD) sensors are used to localize the landmarks as well as to build the maps individually whenever an obstacle is detected. Experimental results show that the robot is capable to effectively determine the position of the landmarks and build a map in a Robotic operating system (ROS).