Research And Design On Vision Orientation Of Care Robot System

With the development of robot intelligence technology, the intelligent servicerobot system is growing up and the robot orientation plays an important role inintelligent robot research, which is the key to realize autonomous navigation andcomplex tasks under certain circumstances. This subject is supported by theresearch projects of Ministry of Education study of Guangdong Province, to thecore of robot visual orientation, searched and collected a large number of domesticand foreign research data and information, analyzed the critical factor that mightaffect the performance of visual orientation, and researched the method of visualorientation in the indoor environment. It is valuable and practically significant forthe robot orientation.First of all, this paper describes the background and significance and theresearch situation of the research, as well as summarizes the characteristics ofvisual positioning. To make sure the reliability and validity of robot orientation, theorientation system structure is designed and the key technical problems areresearched. What’s more, given real-time, high efficiency and general requirementsabout visual information extraction, the method of landmark extraction usinggeometrical invariant moment is proposed after comparative analyzing of severaltypical image segmentation and feature extraction algorithms. Then makeexperiments about the above algorithms on hardware platform, and analyze theerror of the system, which is met the design requirements. Finally, the paperestablishes the model of relevant robot system which combined with indoorenvironment map representation．On the basis of analyzing the method to solve theSLAM problem, the algorithmic framework of the improved EKF-SLAM is raised,which used finite loop differential linearization in filter estimation process. Besides,simulation experiment and real care robot experiment make the algorithm and therobot model together, and the results verify the feasibility and effectiveness of thismethod.The significance of this research is to improve the real-time and reliability ofthe robot visual orientation, and to play a good performance of parallel processingbetween landmark extraction and robot positioning, which is to reduce the timecomplexity of the visual orientation.