A Research On The Reasoning Of Oriented Point Relation Algebra In Three Dimensional Space

Space and time are two problems that human being can not avoid to understanding the world,and are also hot topics in artificial intelligence research.Qualitative spatiotemporal reasoning is an attempt on these two issues.First it abstracts time or space as an object.And then use qualitative descriptive symbols to represent spatial or temporal relationships.Finally,do research on the transformation law between these relationships.Qualitative spatial reasoning models can be used in robot navigation or Wireless Sensor Networks(WSN)to deal with uncertain spatial information.To perform spatial query in a geographic information system(GIS),qualitative spatial relations act as the query predicate,since they are similar to natural language and easy to understand.Qualitative spatial reasoning also has contributed to image retrieval and image classifying,because spatial relations between the regions of interest serve as important features.Directional relationship is one of many spatial relationships,and oriented point relation algebra is a remarkable one in the directional relationship models.Oriented point relation algebra(OPRA)studies the directional relationship between special objects and the reasoning problem between these relationships.It models the space object as a point with a direction,named oriented point.And use the two angles between the conterminous line from one point to another and the directions of the two oriented points,of course,special cases have special trement.Oriented point relation algebra is a qualitative relation model,an angle or angle interval is represented by an integral symbol,which makes it having the following advantages:(1)It is consistent with the general pattern of people's directional cognitive,for example,when walking on the road,people said there is a car in “front”,turning “left” at the crossroads,the “front” and “left” can be regard as angle range symbol.(2)It can extract information from variety of data to establish models,for example,text,voice,camera and so on,certainly,GPS is also possible since it is accurate,this makes the oriented point relation algebra a model that can be used extensively.(3)The directional relationship expressed by OPRA is easy to understand,for example,in two-dimensional space,a model with granularity parameter 1,the direction symbol 0,1,2,3 represent up,left,down,right.(4)It can adapt to different scene with various precision,OPRA has a granularity parameter that can be adjusted,if the simple up-left-down-right can not meet the directional positioning need,one can use the famous clock direction,and the model also provide more other precision.The reasoning problem is an important problem in the relational model.It uses known relations to find unknown relation.Assuming A and B are visible to each other,B and C are also visible to each other.But A and C are invisible to each other for various reasons.Then one can deduce the unknown relation(A C)by using the known relations(A B)and(B C).This is the reasoning problem.Many relation models are so complex that the reasoning on these model is difficult to carry out,such as Pacheco's Integrating 3D Orientation model.But the oriented point relation algebra is simple and the reasoning on it is feasible.In 2012,Mossakowski studied OPRA's composite operations.In 2014,Shengsheng Wang studied OPRA's multi-granularity complex reasoning,which enabled OPRA reasoning to be applied between any mixed granularity.The existing work is mostly carried out in two-dimensional space,involving only three objects,and most of the scene is static.The realistic scene is often threedimensional,often involves more objects,and is often dynamic,which all are calling for new models and reasoning methods.To deal with these problems,this thesis discusse the following questions:(1).The representation model of oriented point in 3-dimension and Oriented Point Relation Algebra in 3-Dimension(OPRA3D)(2).The composition reasoning for OPRA3D(3).Representation and reasoning of dynamic OPRA3 D relations involving multiple objectsFor the representation of the OPRA3 D model,this thesis starts from the modeling of oriented point and gradually builds the whole representation model.For the composition reasoning of OPRA3 D,this thesis gives two geometric constraints of three-dimensional space and its qualitative form,which are used to construct the OPRA3 D composition reasoning algorithm,at the same time,this thesis also discusses the OPRA3 D multi-granularity composition reasoning problem.OPRA3 D also works when facing multi-object dynamic scenes.The OPRA3 D relation network and its sequence are designed to describe the relationship between n objects in this scene.Spatio-temporal reasoning of relation network are proposed to deal with reasoning in multi-object dynamic scenarios.Finally,this paper discusses a problem that needs to be considered when using the OPRA3 D reasoning algorithm in a project,and gives a simulation experiment to verify the effectiveness of OPRA3 D network spatio-temporal reasoning.The modeling method and reasoning algorithm in this thesis have potential value in dealing with the direction relation of objects in three-dimensional space.In the fields of robot navigation,unmanned aerial navigation,space navigation,battlefield analysis and others,these methods are expected to play an important role.