The Importance of the Third Dimension in Robotics
Traditional two-dimensional images are appropriate in many situations. They are excellent for displaying images and for recording, but always subject to interpretation by the human eye. For a computer, analyzing a two-dimensional picture is extremely difficult. Thus, the third dimension is needed. Imaging in 3D significantly simplifies techniques necessary for computer vision. The key to “seeing” objects is not merely to analyse colours and textures, but to segment them in terms of real-world objects and their spatial and temporal relationships in the scene.
With the third dimension, a computer program can distinguish the foreground objects from the background ones. Using this knowledge, the application can interpret its surroundings. Time-of-Flight (ToF) cameras, Stereocameras, RGBD-Cameras produce a depth image, each pixel of which encodes the distance to the corresponding point in the scene. These cameras can be used to estimate 3D structure. There are many practical applications for this new sensing modality, including robot navigation, 3D reconstruction and human-machine interaction.