Research On Dynamic Spatial Information Integrated Multi-aspect Spatial Relations And Application

Because a small advance in qualitative spatial reasoning will bring a great development in those research area which seem the spatial information as its core research contents, which also make the qualitative spatial reasoning has a foundation position in those research area. The actual demand of processing for qualitative spatial information has led the birth and development of the qualitative spatial reasoning.Qualitative spatial information is an important content in a wide variety of areas including robotic navigation, computer vision and geographic information system. As the progess in information technology, the needs of spatial information get higher and higher, and the demand for qualitative spatial information processing has made the research in qualitative spatial reasoning develop rapidly.Qualitative Spatial Reasoning has been an important context in the area of Artificial Intelligence, Spatial information include topology, size, shape, distance etc. Single-aspect spatial information has researched for many years. However, how to combine the single-aspect information in a frame for representation and reasoning is important problem.Most of spatial information need to deal with is integrated multi-aspects spatial relations. These spatial relationships from different aspects use different abstractions to describe, and mutual restraint, mutual influence. Therefore, how to establish a unified framework to represent and reasoning that information is very important. In addition, most of these spatial information need to describe is dynamic. What factors influence the relationships, and how to represent and reasoning about them is very important too.This thesis focuses on the above two problems. We want to build a unified framework to represent these dynamic spatial information that integrated different aspect spatial relations that constraint each other. Moreover, this is very important for qualitative spatial reasoning. In this thesis, several models that integrated multi-aspect spatial relations were proposed, which used to solve the problem Galton proposed. Then a framework for dealing with dynamic qualitative spatial relations are proposed, which is based on the model that can be used in the application of qualitative spatial information, which described in natural language and automated planning.The main contributions in this thesis are as follow:Firstly, using the method of one object with two attributes, a base relation set which integrated with multi-aspects is constructed. Using this method, several modelsb are proposed, which integrated multi-aspects qualitative spatial relationsSecondly, the algorithm of constructing composition table is proposed for reasoning, which made the reasoning result of original model can be still used in new model.Thirdly, to handle the dynamic spatial relations, we proposed the neighborhood partition graph (NPG). In addition, an algorithm is give to generate the GNP. Using this algorithm, one can get the new model's GNP easily. Then,using GNP, we solved the problem which proposed by Galton:"the result of composing two relations in LOS does not always form a conceptual neighborhood".Fourthly, we analyze the model TDSC that integrated topology, size and direction relations. And based on the TDSC model, we propose a framework to deal with qualitative spatial information described in natural language. In addition, we give an example to show the framework is useful. Fifthly, the framework for handing dynamic spatial relations is proposed, which is based on the above models. We use an example to show the framework is correct and effective. In addition, the results demonstrate that this approach is useful in handing that dynamic spatial information which combined multi-aspect relations.In this thesis, we propose a new method which could integrate multi-aspect dynamic spatial information which is better than previous methods. New model can integrate more aspects of spatial informations without changing the resoning algorithm. Neiborhood patition graph is proposed to represent and reason about the qualitative spatial relations and actions. And then the property of neiborhood partition graph is gived. The model TDSC and its neighbourhood partion graph in used for representing and reasoning two problems: natural language description of qualitative spatial information and automated planning with qualitative spatial relations. And some examples is gived to show that our method is correct and effective. This research results in this thesis has some theoretical significance and application value in dynamic spatial information processing.