Research And Implementation Of Semantic Enviroment Modeling And Planning For The Automonous Vehicle

With the development of society and the progress of human civilization, people's requirement and expectation on the robot technology and the products applying such technologies are becoming higher and higher. In order to keep abreast with the trend of social development, practical technologies and ideas, which meet the needs of the new situation and environment better, have to be researched on. It is under this condition that the research task in my thesis is put forward and it is certain that the research in this field is going to have positive influence on the development of this branch of science.Nowadays, the application of robots has been gradually expanded from heavy industry to various areas such as agriculture, medicine, household, service, mining, disaster relieving and space exploring and etc. and this requires not only new ideas but also high-level technologies. Among these challenges, the one facing most scientists and technicians is how to choose the best route, which takes such factors as distance, time, cost and safety into consideration, in a group of many possible ways according to strict estimating criterias and how to deal freely with the emergencies appearing in asymmetrical structures and strange environment while solving the problems effectively in structures and environment already known.Based on the theories of knowledge representation, knowledge base and machine learning in the area of artificial intelligence, on the developing platform of Borland C++ Builder, in this thesis, a relatively detailed and deeper research on the recognition of environment, the build-up of environment and the planning of routes on the semantic level and the learning and precision-making of the features of environment borders in "Multi-environment Practical Robot System" is carried out. Firstly, the semantic level environment modeling of robot is built up according to the three leveled representation mechanism of environmental knowledge. On such a basis, different level tasks are planned after the description, analysis and examination of them. With this being done, the routes ofenvironmental units are organized and produced on the subordination tree of environmental units. As there are virtual environmental units, it is unable to assure the cohesion of the subordination tree's planning. To solve this problem, it is better to search the net through environmental units. In this way, the virtual environment is removed on one hand, and on the other hand, middle areas are inserted so that the environment unit level could be made cohere completely. With regard to the phenomenon that the environmental structure is initially understood but the direction and position of specific border are still unclear and the situation of ignorance of the working area of the robots, the experimental system, by means of analysis, extraction and learning, may understand the environmental structure or make the environment or the borders more precise by tracing the basic environmental data with the help of the information gathered by coder and the photo-electricity switch.In my thesis, firstly, the guiding thought and the method of implementation of the research task are introduced;then the solutions of some crucial problems are given;finally, the design of the system and the results of experiment and test are explicated in details, at the same time, the scientific value of some technologies mentioned in the thesis as well as some limitations and other research potentials are also pointed out.