Visual SLAM with low limit on hardware
Kudan is developing the link between the plethora of devices and sensors that we call the Internet of Things (IoT) and Artificial Intelligence: 3D Recognition Computer Vision Technology. It provides computers the ability to acquire, process, analyse and understand digital images as well as the ability to map its 3D environment, objects, and understand its location within it.
Kudan Visual SLAM Engine: Tech Teaser
Kudan's stereo Visual SLAM mapping and tracking a 6x4m room. Despite being a sparse Visual SLAM solution, a dense point cloud is produced, providing more options for tracking and 3D reconstruction.
Kudan Visual SLAM Engine
Kudan's stereo Visual SLAM mapping and tracking a small desk environment. The stereo baseline is just 3CM and represented by the RGB camera axis length. Sub-millimeter movements are picked up and jitter when the camera remains still is very low.
Kudan Visual SLAM for Headtracking - Tracking an Existing map
Kudan's Visual SLAM system performing monocular tracking a pre-registered map of an office. The camera is held at head height and moved around in fashion consistent with head-tracking use-cases. The camera trajectory remains smooth even under fast motion. By tracking a pre-registered map, there is a lighter load on the processor and tracking times become more constant, allowing very low latency in VR head-tracking applications.