Cybernet’s Computer Vision capabilities have been developed over two decades, and have produced advanced solutions to provide automated recognition of targets, missiles in flight, air/spacecraft outlines and physical road features.
Our systems recognize objects (spacecraft, parts, grasp points, docking targets, or anything that can be defined by a CAD drawing or description) from views taken from one or several cameras, with object descriptions input via commonly used CAD formats like IGES or DXF. When presented with workspace views (including objects and other features such as background clutter, etc.), our vision systems fixate only on objects that they have been “taught.” Once the target has been matched, the object’s position and orientation can be calculated.
Other Cybernet Computer Vision systems are capable of reconstructing the 3D structure of an object from a series of 2D images. The 3D information can then be used to determine the object type, position and orientation. Given enough images, this information provides the necessary data to estimate a missile's trajectory, or allow an autonomously running vehicle to follow the road safely.
Furthermore, Cybernet has developed computer vision-based technology to automate rendezvous and docking of air/spacecraft to support unmanned resupply efforts. This technology seeks to enable autonomous Unmanned Aerial Vehicle (UAV) refueling, effectively extending their maximum mission length. Our approach matches target air/spacecraft images with pre-stored 3D geometry representing the target air/spacecraft, in order to determine misalignment errors in 6 degrees of freedom in real time and correctly insert refueling probes without requiring active human operation.