2022 IB Diploma Extended Essays

The ORB SLAM converts the image to grayscale for its application. The first step is to detect features and initialize the map and its position. Once it gets initialized, it starts creating map. Monocular SLAM requires a procedure to create an initial map because depth cannot be recovered from a single image. One way to solve the problem is to initially track a known structure. In the context of altering approaches, points can be initialized with high uncertainty in depth using an inverse depth parametrization, which hopefully will later converge to their real positions. This makes sparse featured maps such as orb-slam not great at autonomous navigation. This can be also be proven through the graphs above. The trajectory of the robot in relation to the x and y coordinates, implementing ORB SLAM. Which shows the pose of the robot. The robot shows many errors throughout the whole path of the robot. There are many unnecessary movements which is particularly highlighted when the robot is following a straight line or making a turn. Indicative of scale drift at loop closure. Similarly there are many crude changes within the x coordinates in relation to time. Further showing that the change in motion within the x plane isn’t very smooth, the smaller peaks show the robot correcting it’s path in the x plane. What can be seen is that there are many corrections within the x plane and also within the yaw plane demonstrated through yaw(t) normalized.