This talk will give an overview of results obtained during several years of our research in the domain of inertial stabilization and visual tracking for aerial camera platforms. We have been participating in an R&D project coordinated by Czech Air Force and Air Defence Institute which aims at developing a commercializable product. The goal of the platform is to stabilize the line of sight of the optoelectrical payload (day vision camera, infrared camera, laser range finder and laser beamer) in presence of disturbing rotational motions of the carrier (helicopter, unmanned aircraft). On top of the inertial stabilization, automatic image tracking of the ground target is implemented, which compensates for the translational motion of the aircraft and/or the ground object. In this talk we will highlight a few control theoretic problems encountered and solved by our group as the development of the real platform proceeded. All of them were either published in journals or presented at conferences. One such achievement was compensation of nonlinear kinematic coupling between the dynamics in the image plane and the dynamics as seen by the motors in a standard double gimbal system. Another problem was compensation of the one full-sampling-period delay in the slow-sampled outer (visual pointing) feedback loop while being supplied with the measurements from the fast-sampled sensors of inertial angular velocity (MEMS gyros). This compensation invokes the results from reset control systems. Yet another topic reflects the upgrade of mechanical configuration from double gimbal system (with just two rotational axes) to a dual-stage configuration (two joints for each axis), resembling partially the dual-stage control problems known to hard disk drives community. Successful application of structured H-inf design for such system is demonstrated. Finally, feedforward controller from accelerometric measurements is shown as an efficient compensation of disturbance coming through an unbalanced load, reminding us of usefulness of this basic yet often overlooked technique. More info, including papers, photos and videos is at http://aa4cc.dce.fel.cvut.cz/content/inertial-stabilization-and-visual-
Martin Rezac (*1984) is a doctoral student supervised by Zdenek Hurak since 2008. He received his Ing. degree (M.Sc.) in cybernetics and measurement with major in control engineering (summa cum laude) at Czech Technical University in Prague, Prague, Czech Republic, in 2008. He was one of the two key developers of the control system for the inertially stabilized aerial camera platform and his diploma thesis was awarded Dean's prize. His doctoral research is still in the domain of inertial stabilization and estimation in combination with visual servoing. He spent six months in 2009 in Eindhoven University of Technology, The Netherlands, thanks to Erasmus Socrates program. He is a former junior champion of Czech Republic in judo, which confirms his
competitive spirit and ability to focus on what he does.