A synthetic environment for simulation of vision-based formation flight

This paper describes the design of autopilots and formation control laws and then the simulation setup and the first results of a novel architecture for close formation flight based on computer vision. The reference aircraft model is the West Virginia University YF-22 model aircraft. The simulation setup includes aircraft dynamics, autopilots and formation keeping controller and a module that creates a synthetic environment for the simulation of the vision equipment based on a commercial software called Dyna- WORLDS. DynaWORLDS is capable of generating synthetic images as if were captured by the camera onboard the wingman. Uniquely identifiable infrared light markers (light emitters each with different wavelength) are applied to the Leader aircraft and a recent iterative, globally convergent, pose estimation algorithm (LHM) is adopted to reconstruct the leader position and attitude. The formation control laws, designed with GPS measurements in mind, have been applied successfully with the Leader position, velocity and Heading estimates obtained from the vision system alone. A more feasible and less expensive solution using light markers with unique wavelength is then introduced and evaluated yielding the same performance of the previous case.