Research On The Algorithm Optimization Of Autonomous Individual-Soldier Navigation And Drift Reduction Methods

The Individual-Soldier Navigation System using MIMU,can complement the shortages of satellite navigation in indoor areas or tunnels,where satellite signals are weak or lost,and achieve reliable navigation with high accuracy in a period of time.The MIMU-based navigation system can be applied in numerous fields,such as personal localization services,rescue missions as well as tracking and monitoring inside huge buildings.So it has attracted great attention from both academies and industries.This thesis is aimed to satisfy the pratical navigation demands of Individual soldier.This thesis designs Individual –Soldier Navigation System based on micro-sized and low-cost MIMU,and aided with multiple sensors.In order to improve entire accuracy of navigation and localization in comlex walking routes and environments,this thesis proposes several algorithm optimation and drift reduction methods to enable lasting and high accurate navigation in long periods of time.This thesis includes following research works:(1)Research on the Strapdown Inertial Navigation System(SINS)and Pedestrian DeadReckoning(PDR)systematically.Kalman Filter based on the error model ofnavigation system and Zero-Velocity Update(ZUPT)are designed.Throughexperiments,the vality of proposed navigation algorithms is confirmed,and theproperties of both SINS+ZUPT and PDR are evaluated.(2)Based on existed gait detection algorithms,evaluate the Zero-Velocity Detectorsusing accelerator,using gyroscope and fusing both accelerator/ gyroscope.Theexperiments proves that Zero-Velocity Detector fusing both accelerator/gyroscopehas the highest accuracy and throws great influence on navigation sytem.(3)Design activity recognition algorithms using information only from MIMU,whichcan distinguish user's current activity among walking(slow,normal,quick),running,walking on floor/upstairs/downstairs,walking straight/turning.The accivityinformation is applied in other optimation algorithms proposed in this thesis.(4)Design heading drift reduction methods,including reduction methods based onself-contained information from SINS and outside information from other sensors.The reduction methods based on self-contained information include Zero-VelocityUpdate,Zero-Angular Update(ZARU)and Heading Drift Observation Update(HDOU).Experiments prove that the algorithm using ZUPT+ZARU+HDOU hasthe highes accuracy among them.The reduction methods using outside informationinclude the heading provided from magnetor,activity/building heading andcomputer vision.(5)Design height drift reduction methods,including barometer/MIMU-based correctionmethod and Acctivity Recognition/Hidden Markov Model-based correction method.(6)Design Dual-MIMU navigation system.Build dual navigation equations andKalman Filter with inequlity restrains.The experiments illustrate that comparedwith single MIMU navigation system,dual navigation systems have advantages ofhigher localization accuracy,strong reduction of system drifts and faults-tolerantability.(7)Test proposed algorithms through several complex experiments.On MIMU platform,the proposed navigation algorithms obtain a high accuracy with only 0.55%maximum biases in 25 minutes 1000 meters walking.On low-cost MIMU platform,the navigation algorithms can perfom as well.