| 英文摘要 | With the development of the science and technology, the need of high level intelligent mobile robot systems is more and more urgent in many fields. Accordingly, the mobile robotics research comes into a new age. Currently, the capability of environment perception & uncertainty reasoning has become the most important feature of an intelligent mobile robot. Under the support of Chinese 863 High-tech Plan project "Some key technologies of soccer robotics", and the Graduate School of the Chinese Academy of Sciences Innovation project "Intelligent environment perception system for mobile robots", this thesis investigates the problem of environment perception by a mobile robot operating in real world environments. On the basis of uncertainty analysis technology, this thesis focuses on the problem of environment perception, and some related issues are addressed by details, including sensing technology, environment perception & modeling, robot self-localization, real-time navigation & obstacle avoidance. The main work and specific contributions of this research are as follows: Firstly, the application background and development status of mobile robotics are briefly reviewed. Some issues about reasoning with uncertainty in mobile robotics, as well as its scientific context, are addressed. This is followed by details of the research background, structure and main work about the thesis. Secondly, by means of analyzing sensing technology of mobile robotics, an concept of "embedded environment perception system (EEPS)" is presented. Based on the analysis of the potential development of sensing technology, a novel mobile robot oriented EEPS is developed. And then the software & hardware experimental platform in our work is also described in details. Thirdly, due to the fundamental unreliability of the sonar, by investigating the physics of sonar and its working mechanism, a new self-adaptive filtering method for sonar readings is presented, so as to overcome its uncertainty. Then a detailed fuzzy-based sonar sensor model is provided. Experimental results demonstrate the model with high effectiveness and flexibility. Fourthly, by taking into account of the uncertainty of sonar sensors, the discussion turns to the problem of map-building and environment modeling for a mobile robot under real world situations. For the purpose of advancing the performance of sonar, a vision limited sensing (VLS) strategy is employed. To solve the problem of map-building, an approach based on subjective Bayesian reasoning technology is described. To make comparative study, an uncertainty measure D-S evidence reasoning based map-building method is presented. Experimental results show the proposed methods are suitable for sonar-based environment modeling. Fifthly, the problem of adaptively self-localization for a mobile robot in realistic uncertain environments is discussed. The research status of self-localization for mobile robots is |
修改评论