Diz 45 en
Inertial stabilization, estimation and visual servoing for aerial surveillance
Author: Řezáč Martin
This thesis addresses few various sub-topics that are all related together by the topic of inertial stabilization of the optical axis (so called line-of-sight stabilization) for airborne camera systems. In the first part the thesis provides a brief introduction to main principles of the line-of-sight stabilization followed by description of mathematical models for common mechanical configurations such as double-gimbal and dual-stage configurations. Using these models the control algorithms for the line-of-sight stabilization are derived for all mechanical configurations. The second part of the thesis presents the core of author's research achievements related to the topic. Line-of-sight stabilization is extended for a novel image-based pointing-tracking feedback control scheme with real-time computer vision system. The key idea is to enhance the intuitive decoupled controller structure with measurements of the camera inertial angular rate around its optical axis. Since real-time video processing always introduces one sampling period delay in the vision loop, thesis presents several delay compensation schemes based on modified Smith predictor and reset observer approaches. All the algorithms are always implemented in one of the available benchmark systems (inertially stabilized camera platforms) and tested by laboratory or helicopter flight experiments.
- Martin Řezáč, tel: +420 733 114 523, mailto:email@example.com