Implementation of an embedded image-based tracker in a distributed architecture

Abstract

A major purpose of targeting systems is to ease operator workload by keeping the line of sight (LOS) pointed on a designated target. Our airborne electro-optic targeting system uses the latest technology to accomplish this end. If the latitude, longitude and altitude (geo-position) of a stationary ground target are known in advance, this system can use the aircraft's GPS-aided inertial navigation system to keep the LOS pointed at the target. If the operator wants to point at a target whose geo-position is unknown, this system can compute latitude, longitude and altitude using LOS attitude and range. The role of the image-based tracker in this system is to track airborne or moving ground targets and to compensate for drift and other limitations of geo-point for stationary ground targets. The algorithms implemented in our targeting system are not only processing intensive, but also diverse, incorporating a variety of information from numerous sources in order to provide the operator with the best LOS control solution for a given scenario. This, balanced against the low latency requirements of a real-time control system, has resulted in a software design that spans four separate processor boards. We present the implementation of a successful real-time embedded image-based tracker on a distributed architecture.