Visual-Inertial Mapping, Localization and Planning for Autonomous Mobile Robots

Navigation has long been used to determine one's position, location, and path to an object. It provides precise information on the location of a certain place or object. Though there have been several changes and improvements to navigation, there have also been rumours of making it autonomous. This indicates that navigation occurs automatically and without human assistance. The device will understand where it's going and how to get there. Visual odometry (VO) is a key term in this context. It calculates the relative positions of two picture pairs. Similarly, the mobile robot's location is studied. However, there is an issue here. Noise rise (drift error) causes VO to worsen with time. To address this, the Inertial Measurement Unit (IMU) was invented. The IMU consists of accelerometers, gyroscopes, and magnetometers. These produce more accurate findings and aid in noise reduction. The IMU and VO are coupled to generate Visual Inertial Odometry (VIO). An Extended Kalman Filter (EKF) is used to integrate the VIO and GPS for more accurate localization on a local and global scale. Stereo disparity estimation generates a depth sensation map, which is then converted into a 2D occupancy grid map to detect impediments. To avoid obstacles, local path planning generates temporary waypoints, and a waypoint follower directs the device/robot towards the goal.