Underwater Terrain Reference Navigation Base On Topography Feature Fusion

Long-precision real-time navigation is the key Technology for AUV(Autonomous Underwater Vehicle)to complete long-range big deep diving operations,The current inertial navigation system and dead reckoning systems have accumulated error,can not provide long-precision underwater navigation information,Terrain Reference Navigation method does not require an external acoustic arrays support or regular floating to corrected cumulative error,it is significance for AUV submerged for a long time and taking a long-distance cruising.This article will focus on researching the terrain face matching method of underwater Terrain Reference Navigation,All the studding is to found a method that can solve the problem which low acquisition rate of real-time measurement underwater topography,the characteristics in flat terrain regional not obvious,terrain features dispersion and so on.The basic principle of Terrain Reference Navigation is to located the position of a real-time measurement topographic map which the AUV measuring in the pre-stored topographic map using a face matching method.Its essence is the problem of matching and location of three-dimensional surface with big error.Currently;all terrain matching methods are highly value the node sequence comparison methods to find the best similar surface,terrain features in this comparison method and the measurement error significantly smaller area positioning pseudo point does not exist,but the terrain flat areas,or measurement error is large,often making the matching results diverge,there are a number pseudo-anchor point will appear.To solve the problem of locating pseudo point,this paper proposes two matching methods,one is based on Information Fusion method for the node,the node within the terrain surface as a whole,to extract information about the nodes surrounding surfaces,including curvature,gradient,aspect,bump,etc.,more than a single node in the surrounding surface features constitute a set of vectors(packet),not only the node's height value,in order to reduce the amount of calculation using the interval matching method,every n node extract a group of surface features.Actual sea topographic map experiment proved this method has better accuracy and good robustness.Another method is based on the Terrain Partition Matching Positioning Method,the method first calculates the similarity of nodes in matching area,all nodes are divided into identifiable and unidentifiable point,then the matching area will be average partition,The purpose of the block is to partition the point into tow parts,the recognized one and unrecognized one,and the tow part must be apart in maximum extent,The sub-blocks are divided into identifiable and unidentifiable regional area according to the number of identifiable points contained in it,Can identified area with a wealth of terrain features it is helpful to matching location,unrecognized area have flat terrain,the matching position result is vulnerable to the measurement error.assignment a weight values for each sub-block,by using a weight value to reducing effect from the unrecognized area while it taking a positioning process,thereby improving the positioning accuracy and stability,At last taken a simulation experiment using a real underwater topographic map to test the feasibility of this approach,and proves that the method has good positional stability.