

Visual Odometry is the process of estimating the motion of a camera in real-time using successive images. It can also be used for many different applications, ranging from pose estimation, mapping, autonomous navigation to object detection and tracking and many more. As a result, this system is ideal for robots or machines that operate indoors, outdoors or both. Furthermore, one of the most striking advantages of this stereo camera technology is that it can also be used outdoors, where IR interference from sunlight renders structured-light-type sensors like the Kinect inoperable. The 12cm baseline (distance between left and right camera) results in a 0.5-20m range of depth perception, about four times higher than the widespread Kinect Depth sensors. The camera can generate VGA (100Hz) to 2K (15Hz) stereo image streams.

At the same time, it provides high quality 3D point clouds, which can be used to build 3D metric maps of the environment. Advanced computer vision and geometric techniques can use depth perception to accurately estimate the 6DoF pose (x,y,z,roll,pitch,yaw) of the camera and therefore also the pose of the system it is mounted on. In addition to viewing RGB, stereovision also allows the perception of depth. The ZED Stereo Camera developed by STEREOLABS is a camera system based on the concept of human stereovision.
