Research On Bio-Chaotic Information Integration Based On Photoelectric Perception

The research project is the further exploration for the techniques ofphotoelectric information perception and active control under dynamicalenvironments, which are the key criteria to evaluate the levels of intelligence.Additionally, it reflects the future development of a new generation of intelligentphotoelectric sensing and signal processing. Although the machine reasoning hasachieved a lot in the field of artificial intelligence, the abilities ofenvironment-based perception and understanding are in need of improvement.Therefore, the research on environment-based understanding and feedback controlare the hot topics in the field of intelligent photoelectric sensing and signalprocessing. In this dissertation, we present the key techniques of photoelectricinformation integration in perception based on the principle of chaotic dynamics,which provides a new way to solve the problem of robot's active perception andself control. The achievements will be beneficial for further engineeringapplications and academic research. The innovative highlights of this dissertationare shown as follows:1. Considering the theory of Cramer-Rao bound, a novel information infusionand localization technology is presented for multisensor based on optimalgeometrical topology. The optimal sensor-target geometrical localization andtopological characteristics have been studied with three types of photoelectricmeasurements including range-only, time of arrival, bearing-only. The resultsshowed that the sensor-target geometrical topology with optimal localizationperformance could be implemented by analysis of the sensor-target geometricalcharacteristics. The new theory will provide an effective way to explore thegeometrical distribution for multi-sensor network.2. The scheme of robot self-control based on the chaos control method andphotoelectric technology was shown considering not only the physical structure ofthe robot (including the geometric distribution of photoelectric sensors) but alsothe position information of both the obstacles and the targets. The robotphotoelectric sensing-behavior control loop was practically built based on theprinciple of chaotic perception-action dynamical loop. The photoelectric sensing system of the robot was functionally similar to the real perception system in thebrain. The new idea will provide a method for information integration ofphotoelectric sensing and active control.3. In order to solve the problem of self-orientation for motion system undercomplex environments, a novel scheme of robot self-orientation has been shownbased on the assistance of geometric characters and the active perception of therobot. The robot was able to determine the orientation and control the motion allby its photoelectric sensing system. Furthermore, two significant highlights wereconcluded in terms of the effect on the orientation performance of the robot bydifferent geometric characters.4. A new method of positioning was proposed based on the active vision andchaotic evolution. A selection criterion was provided for choosing the set ofmaximal information landmarks in the range of finite field of perception (FOP)based on the information theory and observation entropy. Further, the activeperception model, the motion model and the chaotic perception-action dynamicalloop were established. The experimental tests strongly supported the theory andalgorithms presented in this dissertation and were beneficial for furtherengineering applications.