Design And Research On Visual Positioning And Tracking System For The Small Unmanned Helicopter

Due to the advantages of small size, light weight, low cost, hovering ability, vertical taking off and landing ability, no casualties, and excellent concealment performance, small unmanned helicopter (SUH) has been widely applied in both military and civil areas, for example, military reconnaissance, geological exploration, target tracking, etc. And most applications cannot develop without visual technology; therefore vision-based target positioning and tracking are also key technology in many relevant fields. Compared with those traditional sensors, such as sonar, laser and radar, visual sensor can capture richer and larger amount of natural information, which monitors certain target or context directly.With the experiment platform—SUH, this paper deeply researches on the design and development of an airborne visual positioning and tracking system. According to relevant theories and practical problems encountered during the research, this paper has provided efficient solutions for those chief problems--in terms of target recognition under bad weather conditions, target positioning based on unstable platform, intense vibration of the engine, and so forth. The ultimate goal of this paper is to accomplish the research on hardware establishment, software development, positioning and tracking algorithm design, and flight test.That the target may be partly concealed by exhausts or fog is the main problem in ground target positioning. Therefore, a strong self adaptability of the recognition algorithm is required; moreover, totally auto-recognition needs relatively high accuracy and robust ability. Several theories and algorithms, such as Retinex Theory, Moment Invariants, and Hue-based Segmentation, are combined and improved to achieve our expected result. Then, this paper researches on the HMAX module, which is a bionics vision method based on human brain cortex. This can be the groundwork for the future development of airborne vision system.The main problem in target positioning is to ensure and increase the accuracy of positioning. Since attitude variation and mechanical vibration are two main source of error, a fast correction method is presented to reduce their effect. And a new filter LUKF is designed to filter the intense vibration. LUKF combines advantages of UKF and low-pass filter and can draw a good balance between tracking and smoothing ability. In the end, hardware selection and establishment, as well as software structure and procedure, are discussed in detail. Then some ground experiments, flight tests and their results are shown to verify the visual system. Ultimately, detailed experiment analysis is provided.