Navigable Area Cognition And Path Planning Problem Of Underwater Active SLAM Algorithm

The ability of high-precision underwater navigation and positioning has always been the bottleneck of AUV(autonomous underwater vehicle)to achieve large-scale,long-time,long-distance seabed terrain detection.The push position navigation system can provide continuous underwater pose information for AUV,but its self positioning error accumulates with time,which needs to be corrected by satellite navigation system or acoustic positioning system.However,the coverage of acoustic positioning system is limited,and satellite navigation system can not be used in underwater environment.Although the terrain aided navigation system has high autonomy and does not need satellite navigation system or acoustic positioning system,it needs to preset a priori map,so it is not suitable for AUV to detect the seabed terrain in unknown waters.Simultaneous localization and mapping(SLAM)technology enables AUV to obtain its own pose estimation in the map while building the surrounding environment map in real time.Therefore,slam technology can solve the two problems of AUV positioning and environment map construction at the same time.It has great development potential and space in underwater navigation and positioning technology,and is of great significance to improve the autonomy of AUV.However,slam technology does not have the ability of autonomous cognition of seabed terrain scene,and its intelligence level is low.It can not optimize the navigation path in real time according to the constructed seabed terrain data,and actively correct the positioning error of AUV,which greatly limits the autonomy and intelligence level of AUV.To solve this problem,this paper studies the autonomous cognition and path planning algorithm of underwater active slam technology.Firstly,aiming at the problem that AUV underwater slam does not have the autonomous cognitive ability of seafloor terrain navigability,a nonparametric representation algorithm of seafloor terrain map navigability based on elevation difference is proposed,which is based on DBSCAN(density based spatial clustering of applications with)Noise(DBSCAN)algorithm clusters the nonparametric features of the seabed terrain,and extracts the geographic location information of the navigable area,so that AUV can analyze the elevation map of the seabed terrain,and thus has the ability of autonomous cognition and identification of the environment.Secondly,aiming at the multi beam sonar terrain scanning mechanism,a rotating caliper based on virtual path is proposed Calipers(RC)seabed terrain full coverage path planning strategy enables AUV to quickly perceive the navigable area of seabed terrain,plan the entry mode of navigable area,scan the seabed terrain of navigable area first,and construct high-precision chart of navigable area with arbitrary non concave geometry.Then,aiming at the problem that AUV underwater slam does not have the ability of active correction of positioning error,a path planning algorithm of active correction of AUV positioning error is proposed to construct the active backtracking path of seafloor terrain navigable area.When the navigation system of AUV fails to locate on time,it guides the AUV to avoid obstacles and return to the navigable area for error correction,which improves the autonomy and intelligence of AUV.Finally,aiming at the problem that multi beam sonar can’t match the sensing information of strip seabed terrain sequentially,the soft observation of seabed terrain at the current time is carried out based on Gaussian regression process,and then an AUV positioning error correction algorithm is implemented by using ICP algorithm to match the hard observation and soft observation of seabed terrain,which suppresses the divergence rate of AUV Navigation error.