Validation and Performance Evaluation of Two Different Inertial Navigation System Design Approaches

Similar documents
Spectral Simulation of Turbulence. and Tracking of Small Particles

A Note on the Kou s Continuity Correction Formula

BE.430 Tutorial: Linear Operator Theory and Eigenfunction Expansion

Ideal Amplifier/Attenuator. Memoryless. where k is some real constant. Integrator. System with memory

Optimization-based Alignment for Strapdown Inertial Navigation System: Comparison and Extension

Introduction to Mobile Robotics Mapping with Known Poses

Big O Notation for Time Complexity of Algorithms

Paper Introduction. ~ Modelling the Uncertainty in Recovering Articulation from Acoustics ~ Korin Richmond, Simon King, and Paul Taylor.

Fresnel Dragging Explained

April 1980 TR/96. Extrapolation techniques for first order hyperbolic partial differential equations. E.H. Twizell

Section 8 Convolution and Deconvolution

EGR 544 Communication Theory

A MULTI-OBJECTIVE MODEL FOR DECISION MAKING IN A CLOSED-LOOP SUPPLY CHAIN NETWORK

SUMMATION OF INFINITE SERIES REVISITED

Optimum design of complementary transient experiments for estimating thermal properties

State-Space Model. In general, the dynamic equations of a lumped-parameter continuous system may be represented by

In this section we will study periodic signals in terms of their frequency f t is said to be periodic if (4.1)

Nonlinearity-Compensation Extended Kalman Filter for Handling Unexpected Measurement Uncertainty in Process Tomography

CSE 5311 Notes 12: Matrices

Economics 8723 Macroeconomic Theory Problem Set 2 Professor Sanjay Chugh Spring 2017

N! AND THE GAMMA FUNCTION

ES 330 Electronics II Homework 03 (Fall 2017 Due Wednesday, September 20, 2017)

[ m] x = 0.25cos 20 t sin 20 t m

ARAR Algorithm in Forecasting Electricity Load Demand in Malaysia

Comparison between Fourier and Corrected Fourier Series Methods

Comparisons Between RV, ARV and WRV

Behaviors of MIMO UWB-IR Transceiver with Statistical Models

( F, ρ )-Convexity in Nonsmooth Vector Optimization over Cones

Notes 03 largely plagiarized by %khc

12 Getting Started With Fourier Analysis

Calculus Limits. Limit of a function.. 1. One-Sided Limits...1. Infinite limits 2. Vertical Asymptotes...3. Calculating Limits Using the Limit Laws.

1. Solve by the method of undetermined coefficients and by the method of variation of parameters. (4)

Integrality Gap for the Knapsack Problem

Optimization of Rotating Machines Vibrations Limits by the Spring - Mass System Analysis

The analysis of the method on the one variable function s limit Ke Wu

Electrical Engineering Department Network Lab.

BEST LINEAR FORECASTS VS. BEST POSSIBLE FORECASTS

Energy Density / Energy Flux / Total Energy in 1D. Key Mathematics: density, flux, and the continuity equation.

Sliding Mode Control for Robust Stabilization of Uncertain Input-Delay Systems

6.01: Introduction to EECS I Lecture 3 February 15, 2011

Chapter 3 Moments of a Distribution

A Note on Random k-sat for Moderately Growing k

Calculus BC 2015 Scoring Guidelines

Let s express the absorption of radiation by dipoles as a dipole correlation function.

Department of Statistics, College of Science, Persian Gulf University, Bushehr, Iran

Stationarity and Error Correction

The Importance of Ordering the Number of Lattice Points Inside a Rational Polyhedron Using Generating Functions

Chemistry 1B, Fall 2016 Topics 21-22

1 Notes on Little s Law (l = λw)

Supplementary Figure S1 Characterization of the two-qubit system. a-d, When the nuclear spin is polarized along the

INVESTMENT PROJECT EFFICIENCY EVALUATION

Zhi-Wei Sun and Hao Pan (Nanjing)

10.3 Autocorrelation Function of Ergodic RP 10.4 Power Spectral Density of Ergodic RP 10.5 Normal RP (Gaussian RP)

A Generalized Cost Malmquist Index to the Productivities of Units with Negative Data in DEA

Actuarial Society of India

Improvement Over General And Wider Class of Estimators Using Ranked Set Sampling

Effect of Heat Exchangers Connection on Effectiveness

Available online at ScienceDirect. Procedia Computer Science 103 (2017 ) 67 74

2 f(x) dx = 1, 0. 2f(x 1) dx d) 1 4t t6 t. t 2 dt i)

Some Properties of Semi-E-Convex Function and Semi-E-Convex Programming*

L-functions and Class Numbers

State and Parameter Estimation of The Lorenz System In Existence of Colored Noise

WILD Estimating Abundance for Closed Populations with Mark-Recapture Methods

MITPress NewMath.cls LAT E X Book Style Size: 7x9 September 27, :04am. Contents

CS623: Introduction to Computing with Neural Nets (lecture-10) Pushpak Bhattacharyya Computer Science and Engineering Department IIT Bombay

Adaptive Visual Servoing of Micro Aerial Vehicle with Switched System Model for Obstacle Avoidance

Sheffer sequences of polynomials and their applications

Solutions to selected problems from the midterm exam Math 222 Winter 2015

Processamento Digital de Sinal

FIXED FUZZY POINT THEOREMS IN FUZZY METRIC SPACE

The Eigen Function of Linear Systems

Extended Laguerre Polynomials

Solutions Manual 4.1. nonlinear. 4.2 The Fourier Series is: and the fundamental frequency is ω 2π

ODEs II, Supplement to Lectures 6 & 7: The Jordan Normal Form: Solving Autonomous, Homogeneous Linear Systems. April 2, 2003

On Acoustic Radiation by Rotating Dipole Sources in Frequency Domain

Bernstein Direct Method for Solving. Variational Problems

TRANSIENT PROCESES FOR VENT-EJECTORS ASSISTED BY COANDA EFFECT

A quadratic convergence method for the management equilibrium model

Using Linnik's Identity to Approximate the Prime Counting Function with the Logarithmic Integral

Problems and Solutions for Section 3.2 (3.15 through 3.25)

Dynamic Model and Analysis of Nonlinear Vibration Characteristic of a Curve-Face Gear Drive Cai, Z. Lin, C. Zhiqin Cai Chao Lin*

Class 36. Thin-film interference. Thin Film Interference. Thin Film Interference. Thin-film interference

EXISTENCE THEORY OF RANDOM DIFFERENTIAL EQUATIONS D. S. Palimkar

A TAUBERIAN THEOREM FOR THE WEIGHTED MEAN METHOD OF SUMMABILITY

Fuzzy PID Iterative learning control for a class of Nonlinear Systems with Arbitrary Initial Value Xiaohong Hao and Dongjiang Wang

King Fahd University of Petroleum & Minerals Computer Engineering g Dept

Math 2414 Homework Set 7 Solutions 10 Points

A Probabilistic Nearest Neighbor Filter for m Validated Measurements.

b : the two eigenvectors of the Grover iteration Quantum counting algorithm

IMPROVED VEHICLE PARAMETER ESTIMATION USING SENSOR FUSION BY KALMAN FILTERING

E will be denoted by n

MODELING THE DYNAMICS OF A MONOCYCLIC CELL AGGREGATION SYSTEM

Discrete-Time Signals and Systems. Introduction to Digital Signal Processing. Independent Variable. What is a Signal? What is a System?

Supplementary Information for Thermal Noises in an Aqueous Quadrupole Micro- and Nano-Trap

CLOSED FORM EVALUATION OF RESTRICTED SUMS CONTAINING SQUARES OF FIBONOMIAL COEFFICIENTS

CHARACTERIZATIONS OF THE NON-UNIFORM IN TIME ISS PROPERTY AND APPLICATIONS

Available online at J. Math. Comput. Sci. 4 (2014), No. 4, ISSN:

Math 6710, Fall 2016 Final Exam Solutions

Dynamic h-index: the Hirsch index in function of time

STK4080/9080 Survival and event history analysis

Transcription:

Ieraioa Goba avigaio Saeie Syses Sociey IGSS Syposiu 7 The Uiversiy of ew Souh Waes, Sydey, Ausraia 4 6 Deceber, 7 Vaidaio ad Perforace Evauaio of Two Differe Ieria avigaio Syse Desig Approaches Di i Depare of Eecrica Egieerig / Écoe de Techoogie Supérieure (ETS) / Uiversiy du Quebec / aada Phoe: + ( 54 ) 396-88 ex. 7558 / Fax: + ( 54 ) 396-8684 / Di.i@es.ca Reé Jr. adry Depare of Eecrica Egieerig / Écoe de Techoogie Supérieure (ETS) / Uiversiy du Quebec / aada Phoe: + ( 54 ) 396-88 ex. 7558 / Fax: + ( 54 ) 396-8684 / radry@ee.es.ca Phiippe avoie Depare of Eecrica Egieerig / Écoe de Techoogie Supérieure (ETS) / Uiversiy du Quebec / aada Phoe: + ( 54 ) 396-88 ex. 7558 / Fax: + ( 54 ) 396-8684 / phiippe.avoie.5@es.es.ca ASTRAT For ay years, he desig ad ipeeaio of a pracica IS agorih have bee a subjec of grea ieres i cosideraio of cerai issues, such as he copuig oad, precisio, speed/updaig rae ad copuer eory. I he as decade, as he rapid deveope of copuer hardware/sofware echoogy, he cocer of he copuig copeecy is o criica ayore; herefore he cassic sophisicaed IS desig ehod ay be sipified by akig he advaages of oder copuer s capabiiies. This sudy is o ipee ad evauae wo differe IS desig approaches. Specificay for he cassic wo-speed digia IS desig, he discree IS odes, he high/ow speed digia iegraio agorihs, coig/scuig/scroig copesaios for he ow speed cacuaio i aiude/veociy/high precisio posiioig deeriaio are ipeeed i a prograig eviroe. For he sipified IS desig, he sige high-speed IS agorih free of coig/scuig/scroig copesaios is

ivesigaed. Moreover by uiisig Maab Siuik s capabiiy o sove he coiuous-ode differeia equaios, isead of usig he discree IS odes, he coiuous IS odes are direcy epoyed i he sipified IS desig. The perforaces of he wo deveoped IS desigs are vaidaed ad evauaed iside a iegraed GPS/IS souios based o a pracica oosey-couped Kaa fier. Rea-ie IMU raw easurees ogged fro he acic-grade Rig aser Gyros (RG)/acceeroeers ad GPS souios correspodig o he road esig rajecory is uiised i he vaidaio ad evauaio. The copariso of he wo IS souios shows ha, boh of he IS desigs reach o he sae eve of precisio wih respec o he provided refereces, however he processig ie of he cassic wo-speed digia-ode IS is reduced 9% copared wih sige speed coiuous-ode IS. However copared wih he cassic wo-speed IS approach, i er of he agorih/sofware copexiy, i is reduced by 5-6% for he sige speed IS which wi peri he fas-prooypig ad easy-o-udersad desig. Moreover a pracica Kaa fier was deveoped by iegraig IS wih he GPS posiio ad veociy daa o be abe o deiver he accurae ad robus avigaioa souios. KEYWORDS: IS, Kaa fier, wo-speed IS, IMU. ITRODUTIO Ieria avigaio is a auooous process of copuig posiio ocaio by douby iegraig he acceeraio of a poi whose posiio is o be deeried. The fudaea pricipe is depiced i Figure. Figure. asic cocep of Ieria avigaio Syses A srapdow Ieria avigaio Syse (IS), i which he Ieria Measuree Ui (IMU) physicay oued o he vehice, ipees he Figure. pricipe usig he riad of aguar roaio sesors, e.g. gyros, o deerie he direcio of he acceeraio vecor sesed by he riad of he acceeroeers. y perforig he iegraio i a digia copuer, he veociy ad posiio of he vehice ca be deeried. I geera, oa acceeraio coprises wo pars: graviy acceeraio ad specific force acceeraio. Due o he iiaio of oder echoogy, he acceeroeers oy easure he specific force copoe of he acceeraio. Therefore, o derive he oa acceeraio, he graviy acceeraio us be cacuaed ad added o he acceeroeer easuree.

Theoreicay, he aiude/veociy/posiio cacuaios are aayicay described i he for of coiuous differeia equaios ha whe iegraed i he cassica aayica coiuous sese, woud provide he aiude, veociy ad posiio daa. Whie i a acua IS, he agorih suiabe for he IS copuer shoud be desiged o achieve he sae uerica resus by digia iegraio operaig a a specified repeiio rae as he coiuous iegraio of he differeia equaios. Sice he basic srapdow IS cocep cae io beig, he researchers have aiy focused o he desig of agorih for he aiude iegraio, which is because ha eeig specified aguar rae sesor accuracy requirees uder high dyaic aguar rae ipus is geeray ore difficu o achieve (Savage P.G. ), whe he srapdow aguar sesors direcy are exposed o he high dyaics of he vehice. Due o he capabiiies ad iiaios of he copuer echoogy, he srapdow IS uses aiude agorih based o a wo-speed approach. The high speed approach wih s order agorih is o accuraey accou for he high frequecy aguar rae osciaio ha ca recify by providig a syseaic aiude buid-up, he oderae speed higher order agorih accuraey accous for arger apiude aguar oio. Take ogeher, he cobied accuracy of wo speed approach is equivae o operaio he higher order agorih a high speed rae, however wih ower copuig hroughpu. ased o he sae fraework, a wo speed approach ca aso be used for acceeraio rasforaio/veociy iegraio i a dyaic eviroe ha paraes he wo speed aiude iegraio approach. The high speed agorih is o accou for high frequecy aguar ad iear osciaio ha ca recify io syseaic veociy buid-up, he ow speed agorih perfors he acceeraio rasforaio based o ipus fro he high speed agorih. For he posiio iegraio, he sipe rapezoida iegraio of veociy a a updae rae equa o or ower ha he veociy updae frequecy is ypicay epoyed i ay oder srapdow IS copuer as we i his sudy. I high posiioig appicaio, he wo speed fraework is aso used o deveop a high precisio posiio iegraio agorih (Savage P.G. ). Whie he ie of wo speed approach was o overcoe he hroughpu iiaio of eary copuer echoogy, ha iiaio is rapidy becoig isigifica wih coiuousy rapid advaces i oder high speed copuer echoogy. This provides he oives o desig a siper sige speed agorih srucure i which a copuaios are execued a a repeiio rae ha is sufficiey high o accuraey accou for high frequecy aguar rae ad acceeraio recificaio effecs. The purpose of his sudy is o evauae he perforace of he cassic wo speed IS digia iegraio agorih desig by coparig wih a sipified sige high speed IS iegraio agorih desig. Firs of a, he cassic wo speed IS digia iegraio agorih icudig he coig, scuig ad scroig copesaio respecivey for he aiude, veociy ad posiio cacuaio is ipeeed i prograig eviroe. Secody, a siper sige high speed IS agorih srucure based o he IS aayicay coiuous differeia equaios is ipeeed, by uiisig Maab Siuik s capabiiies o direcy sove he differeia equaios i he coiuous ode as we he high copeece of oder copuer echoogy. Fiay i order o obai a robus IS souio, a pracica oosey-couped iegraed IS/GPS Kaa fier is desiged o esiae he IMU sesor rado oises as we o overcoe he ie-accuuaig IS daa errors by opiay coupig GPS posiioig ad veociy daa. I he experiea resus, he rea-ie ogged acic grade IMU raw easurees are pos-processed by he wo IS desigs ad he iegraed wih GPS daa by he proposed Kaa fier; he road es rajecory is ieioay aoeuvred o represe differe dyaics. The IS daa of he wo desigs is copared i ers of precisio, cacuaig ie ad sofware copexiy. The resus show

ha boh of he IS desigs reach o he sae eve of precisio wih respec o he provided refereces. Fiay I he iegraio of IS/GPS es, he avigaio souios i ers of aiude, veociy ad posiio are greay ehaced by he GPS daa.. ASI MATHEMATIA AKGROUD IS uses acceeraio ad aguar rae easurees refereced o ieria space, bu aiude, veociy ad posiio are refereced are coveioay eeded i a syse refereced o he earh. osequey he foowig coordiae fraes defied hroughou he paper are iroduced o evauae he IMU easurees (Savage P.G. ). E Frae Earh fixed frae for posiio ocaio defiiio, Y axis parae o he earh poar axis, X axis passes hrough he Greewich eridia whie paraes o he equaoria pae, Z axis is perpedicuar o X ad Y axis. Frae avigaio frae wih Z axis parae o he upward verica a he oca earh surface, used for iegraig acceeraio io veociy ad for defiig he aguar orieaio. Frae ocay eve frae parae o he frae bu wih Z axis parae o he dowward verica, ad X, Y axes aog frae Y, X axes, used for describig he srapdow sesor frae orieaio. Geo Frae ocay eve geographic frae defied wih is Z axis upward aog he oca geodeic verica, Y axis orh wih X axis eas. Frae Vehice fixed frae wih axes parae o righ haded orhogoa sesor ipu axes. Respecivey, X axis aog he righ wig of he vehice, Y axis aog he ogiudia axis of he vehice ad Z axis dowward perpedicuar o X, Y axes. I Frae Absoue o-roaig ieria frae used for a referece for aguar roaio easurees, Y axis is direced owards he ceesia orh, X axis owards he vera equiox ad Z axis perpedicuar o he X, Y axes. eed o eio, a he coordiae fraes X, Y, Z axes foow he righ haded orhogoa rues. 3. DIGITA STRAPDOW IS ITEGRATIO AGORITHM IMPEMETATIO I a acua IS copuer, as i is aforeeioed, he iegraio fucios obaiig he aiude, veociy ad posiio are execued wih digia agorihs operaig a a specified repeiio rae. I geeray icudes foowig seps, firsy iegraio of IS aguar rae sesor daa io aiude Aiude Iegraio, secody usig aiude daa o rasfor IS acceeroeer daa fro frae i frae Veociy Iegraio ad fiay iegraio of frae veociy io posiio Posiio Iegraio. As a exape of he wo speed digia iegraio fraework ipeeig Aiude, Veociy ad Posiio Iegraio, he wo speed agorih of Aiude Iegraio is expaied ) i deais i Figure. is he aiude arix describig he reaio bewee Frae o )

Frae o be deeried. A sipe high speed s order agorih accous for sa apiude high frequecy aguar osciaios ha ca recify io syseaic aiude buid-up i ers of oig copesaio β i oop ( - 4 KHz ). Depedig o owadays ieria sesor echoogy, oher ha direcy providig he aguar rae, os of aguar rae sesor oupu he iegraed aguar rae icrees Δ α over a cerai sapig period. y akig he oupus fro high speed oop, a oderae speed higher order agorih accuraey accou for arger apiude aguar oio over he oderae speed updae cyce period i oop ( 5 Hz ). Frae aiude chage fro epoch o epoch - ) is achieved by sovig he roaio vecor differeia equaio & φ ) i digia approxiaioφ. ) ) = ) ) ) ) ) ) = ) ) ) ) ) si ξ ( cosξ ) ) = I ( ξ ) + ( ξ )( ) ξ ξ ξ ) siφ ( cosφ) ) = I + ( φ ) + ( φ )( ) φ φ φ I ω ξ & = I ω d [ ω IE / + ρ u Z + F ( u Z v )] Z / / j ξ ω IE / T + ρz u ZT + ( Δ ) / F u / Z R M M ΔR ( v + v ) T & φ siφ φ ( ) ( ) ( = ω I + φ ω I + φ φ φ ω I ) ( cosφ) α β φ = α + β α ( ) = ( ) ω Idτ α = α β = I d ( α( ) ω ) τ oig = α + Δα = α = α ( = ) α = ( ) = β + Δβ β = β ( = ) β = ( = ) - - Δβ = ( α Δα ) + ( Δα ( ) ωi ) d τ Δα = ωidτ Δβ = ( α + Δα ) Δ 6 α s order - Figure. Schee of Aiude acuaio To accopish he aiude cacuaio depiced i Figure, Frae aiude chage caused by he vehice rasaioa oio eeds o be updaed. Due o sa chages i he veociy ad posiio over he cacuaio ierva, he Frae roaio vecor ξ is very sa. Hece i ca be cacuaed by approxiaig a o-roaio Frae rae vecor ω I i he ow speed updae cyce period i oop ( 5 Hz ). ased o he sae wo speed fraework, he veociy iegraio is depiced i he Figure 3. ) Frae aiude chage fro epoch o epoch - is used o deerie he direcio ) ) he acceeraio vecor. osequey he cacuaio of coaiig he oop rae ) oig copesaio is icuded i he veociy iegraio. Whe coupig wih aiude

oig effec, he acceeraio vecor geeraes iear veociy osciaio. Therefore a high speed s order agorih accous for high frequecy aguar/iear osciaios ha ca recify io syseaic veociy buid-up Scuig copesaio i oop ( - 4 KHz ). A oderae speed higher order agorih perfors he acceeraio rasforaio based o ipus fro he high speed agorih i oop ( 5 Hz ). Meawhie i oop, he veociy caused by he graviy ad oriois acceeraio shoud be cacuaed. ecause graviy ad oriois acceeraio is a fucio of he vehice posiio ad earh roaio/rasaioa rae respecivey, i ca be cacuaed by approxiaig a o-roaio Frae rae vecor I ω i he ow speed updae cyce period i oop ( 5 Hz ). V = V + SF + G / OR G / OR = [ g ( ω E + ω IE ) V ] d P [ ω + ρ u + F ( u v ) v ]} T G / OR { g P IE Z Z / / / / Z / / SF = ) SF SF = = ) ) a SF d ) ) ) ) a SF d Δ V SF = ) ) Δ V ) SF ) ) ) SF = ) SF ) SF = ) ) a SF d Figure 3. Veociy Iegraio I he posiio deeriaio, he sipe rapezoida iegraio as we he high resouio posiio agorih coaiig high speed Scroig copesaio is ipeeed o accou for he dyaic aiude ad veociy chages wihi he posiio updae period. 4. OTIUOUS STRAPDOW IS ITEGRATIO AGORITHM IMPEMETATIO The sae basis as he digia IS agorih, heoreicay, he aiude/veociy/posiio cacuaios are aayicay described i he for of coiuous differeia equaios ha whe iegraed i he cassica aayica coiuous sese, woud provide he aiude, veociy ad posiio daa. I his sudy, by uiisig Maab Siuik s capabiiies o direcy sove he differeia equaios i coiuous ode ad he high copeece of oder copuer echoogy, a siper sige high speed IS agorih srucure based o he IS aayicay coiuous ie-rae differeia equaios is ipeeed as we. The aiude is achieved by he iegraio of he aiude ie rae equaio give by: & = ( ωi ) ( ωi ), ω I = ( ωie + ωe ) () Where, Aiude arix, ω I skew-syeric arix of he aguar rae vecor i frae, ωi skew-syeric arix of he aguar rae vecor caused by he rasaioa oio i frae, ω E he aguar rae of Frae reaive o E Frae, i.e. so-caed raspor rae, ω IE earh roaio rae. The Maab ipeeaio of aiude iegraio is give by Figure 4.

Figure 4. Aiude Iegraio The veociy is deeried by he iegraio of he veociy ie rae equaio give by: v & = asf + gp ( ωe + ω IE ) v, gp = g IE ( ωie R) ω () Where, v is veociy vecor, a SF specific force vecor, g P Pub-bob graviy, g is Sadard graviy, R is posiio ocaio vecor fro he earh ceer. The Maab Siuik ipeeaios of veociy iegraio ad raspor rae are give i Figure 5 ad 6 respecivey. Figure 5. Veociy Iegraio The posiio deeriaio ca be accopished by iegraig he foow equaio: & E = E ( ω ) E & (3) h = v Where, ω E he skew syeric arix of raspor rae, E Direcio osie Marix (DM) fro frae o E frae, i.e. he so-caed posiio arix, v verica copoe of veociy vecor.

Figure 5. Ipeeaio of Traspor Rae 5. ITEGRATED KAMA FITER IMPEMETATIO IS sad-aoe souio has iied by is ie-growig errors due o he iperfecio of ieria sesors, i.e. ieria sesor bias, rado oise, scae facor isabiiy ec. I he copariso wih IS, Goba avigaio Saeie Syse (GSS), such as GPS, GOASS ad Gaieo, is capabe of deiverig posiio ad veociy iforaio asceraied wih ieidepede precisio. Whie because GSS is vuerabe o he eviroe ierferece, such as radio eissio, buidig bockage of he siga ec, herefore as a oay auooous avigaio souio, IS is cooy couped wih GSS by Kaa fier echique o auge he over-a perforace. Geeray he desig of Kaa fier error sae rasiio arix is criica o he iegraio syse perforace. The rasiio arix desiged i his sudy is o he basis of IS dyaic error odes which are give by (Savage P.G.): Ψ& δv& δ R& = δω = δ a = δv I SF ω a E I ω Ψ SF Ψ δ R + δ g Md IE (ω + ω E ) δv (4) Where, Ψ,δ V, δ R Aiude, veociy ad posiio error paraeers, DM fro o frae, δ g Pup-bob graviy error, ω avigaio frae roaio rae wih respec o he Md I ieria frae, by defiiio which equas o ω IE + ω, he defiiios of δ ω E I are sae o he Eqa. ()-(4). I soe docueaio, he equaios (3) refer o IS ψ -age error ode, i which he IS error paraeers, i.e. aiude/veociy/posiio errors are defied i he E Frae ad expressed i Frae. The Kaa fier error sae rasiio arix ca be achieved by discreizig he ie-rae differeia Eqa. (4) wih he proper agorih, i his sudy d order agorih is uiised:

Φ = I + ΔΦ λλ + ΔΦ λλ ΔΦ λλ A( ) T = A( ) d = [ A( ) + A( )] T T = (5) Where, A() is he coiuous-ode IS dyaic error ode represeed by Eqa(4), is he ie ierva beig used o rasi he IS error saes bewee wo epochs, e.g. i his sudy, i is he sae as he cyce updae rae, i.e. Hz. ecause he ie ierva T is sa, A() paraers ca be reaed as cosas or sowy chagig variabes i he iegraio durig he T ierva. For isace, he Maab Siuik ipeeaio of IS dyaic error ode ad he discreizaio is give by Figure 6. Figure 6. Siuik ipeeaio of Kaa fier error sae rasiio arix 6. SYSTEM VAIDATIO AD PERFORMAE EVAUATIO The road es was coduced i a parkig o i ava, Québec o he 4 h Juy 7. I he experiea resus, he rea-ie ogged acic grade IMU raw easurees are posprocessed by he wo IS desigs ad he iegraed wih GPS daa by he proposed Kaa fier; he road es rajecory is ieioay aoeuvred o represe differe dyaics. I he es, IMU Equipe Hoeywe 7 AG58 i Figure 7(a) was uiised, which specificaios are provided by Tabe. The GPS receiver used i he es is avae D-4 i Figure 7(b). The es asseby is depiced i Figure 7(c)/(d). The avae iegraed GPS/IS souios es IS PVA were epoyed i he es as he referece. Tabe. Specificaios of Hoeywe 7 AG58 Ipu Rage (deg/sec) Rae ias (deg/hr) Gyro Rae Scae Facor Aguar Rado Wak (deg/r-hr) Ipu Rage (g) ias (g) Acceeroeer Scae Facor Acceeraio Rado Wak (ug/r-hz)

./ (858993459.).5 5.348/ (34778.) 34 I he es, firs of a he sad-aoe IS souios respecivey fro ad Siuik ipeeaios are copared wih he referece. The sad-aoe IS souios are obaied by processig he raw IMU daa, i.e. riad specific forces/aguar raes. Figure 8- (a) depics he IS sad-aoe souios fro Siuik approach. The es is for 3s, he processig ie is abou 6s. y processig he sae raw IMU daa, he IS sad-aoe souios by approach are obaied ad depiced i Figure 8- (b). The processig ie by IS sofware is abou 4-5s. oparig he processig ie, obviousy approach processes uch faser ha Siuik, which akes progra is ideay for a rea ie IS souio ha Siuik. (a) (b) (c) (d) Figure 7. Experiea es equipes ad seup y coparig wih ovae bes PVA, depiced i Figure 9-, he sad-aoe IS aiude, veociy ad posiio souio fro /Siuik approach reach he sae eve precisio. I ca be foud ceary ha he sad-aoe IS has he ie-depede errors, which are aiy caused by he IMU sesor bias ad oise. I aso ca be cocuded ha boh of he Siuik ad IS approaches deiver he sad-aoe IS souios wih he sae accuracy, whie progra is uch faser ha Siuik i er of processig ie.

Siuik Souio ovae Souio Souio ovae Souio og (rad) og (rad).7954.7954.7955.7955.7955.7955.7955 a (rad).7954.7954.7955.7955.7955.7955.7955 a (rad) (a) (b) Figure 8. IS sad-aoe souio fro Siuik ad vs. ovae es Posiio souio Age (rad) Age (rad) Age (rad) Siuik Souio Ro (rad). ovae Souio -. Pich (rad).5 Headig (rad) - Age (rad) Age (rad). Ro (rad) Souio ovae Souio -. Pich (rad) Age (rad).5 Headig (rad) 5 (a) (b) Figure 9. IS sad-aoe souio fro Siuik ad vs. ovae es Aiude souio eed o eio ha i Figure 9 (a), he big age headig differece bewee Siuik souio ad he referece is π. Ve (/s) Ve (/s) Ve (/s) Siuik Souio Ve Eas (/s) ovae Souio - Ve orh (/s) - Ve UP (/s) - Ve (/s) Ve (/s) Ve (/s) Ve Eas (/s) Souio ovae Souio - Ve orh (/s) - Ve UP (/s) - (a) (b) Figure. IS sad-aoe souio fro Siuik ad vs. ovae es Veociy souio

og (rad) a (rad).7955.7954 a (rad) Siuik Souio ovae Souio.7954 og (rad) -.864 A () A () - og (rad) a (rad).7955.7954 a (rad) Souio ovae Souio.7954 og (rad) -.864 A () A () - (a) (b) Figure. IS sad-aoe souio fro Siuik ad vs. ovae es Posiio souio Due o he IMU sesor errors, he IS sad-aoe souio has he ie-growig errors. To esiae ad copesae hose errors, referrig o he aforeeioed reasos, he Kaa fier-based IS/GPS iegraio is he idea souio o deiver he opia aiude, veociy ad posiio. Moreover accordig o he basic pricipes of IS, he verica chae is auray diverge i he IS sad-aoe cacuaio, which ca be proved by Figure (a)/(b) refer o aiude souios. To ake he aiude cacuaio coverge, he exera correcio for he aiude coro is required. y uiisig he accurae GPS aiude iforaio, he aiude coro of IS ca aso be achieved by he iegraio of IS/GPS. Figure depics he iegraed IS/GPS souio copared wih he referece. Souio ovae Souio Siuik Souio ovae Souio og (rad) og (rad).7954.7954.7955.7955.7955.7955 a (rad).7954.7954.7955.7955.7955.7955 a (rad) (a) (b) Figure. IS/GPS iegraio souio fro Siuik/ vs. ovae es Posiio souio I he es, accordig o he referece, he disace bewee each sraigh-ie rajecory is abou 3 eers. Therefore he errors bewee iegraed IS/GPS souio ad he referece is roughy abou.5 eer, which ca be see fro Figure. Figure 3-5 show he errors of Siuik/ iegraio souio copared wih ovae es PVA. Fro he resus, i ca be observed ha he approach based IS/GPS iegraio has saer ad ess oisy errors ha Siuik-based IS/GPS iegraio souios, which is because he digia iegraio agorih used i IS/GPS iegraio are ore accurae ad we desiged specified for he IS cacuaio ha he sipified coiuous-ode iegraio agorih epoyed i Siuik IS/GPS agorih.

Ve error (/s) 5 Eas Ve Error (Siuik vs ) Siuik Souio Souio -5 4 orh Ve Error (Siuik vs ) Ve error (/s) Ve error (/s) - -4 UP Ve Error (Siuik vs ) - -4 Figure 3. IS/GPS iegraio veociy errors of Siuik/ copared wih ovae es Veociy souio Age error (rad) 5 Ro Error (Siuik vs ) Siuik Souio Souio -5 Age error (rad) Age error (rad) Pich Error (Siuik vs ) - - Headig Error(Siuik vs ) 5-5 - Figure 4. IS/GPS iegraio veociy errors of Siuik/ copared wih ovae es Aiude souio I is he sae as i Figure 9, he big headig errors is π adjuse, where π ad idicae he sae headig, i.e. rue orh i Figure 4.

a error (rad) 5 x -6 a Error (Siuik vs ) Siuik Souio Souio -5 og error (rad) x -6 og Error (Siuik vs ) - - A Error (Siuik vs ) 5 A error () 4 3 Figure 5. IS/GPS iegraio veociy errors of Siuik/ copared wih ovae es Posiio souio 6. OUSIO This paper focuses o he desig of he IS agorih. Two IS approaches respecivey i prograig-based digia ode ad Siuik-based coiuous ode are ipeeed. Rea daa road es vaidaes he boh proposed desigs. Firsy i he sad-aoe IS es, wih fairy big IMU sesor syseaic bias ad oises, however he sad-aoe IS souios fro ad Siuik approaches foow he referece rajecory. This vaidaes he sadaoe IS souios by /Siuik ipeeaio. Wih he copariso of ad Siuik IS, boh of he approaches reach he sae eve of precisio, everheess he prograig-based sad-aoe IS is iproved 9% i er of he processig ie cosupio copared wih Siuik-based oe Moreover he foowed vaidaio es shows ha Kaa fier is abe o effecivey esiae/copesae he raw IMU rado oises by uiisig he aid of he GPS posiio ad veociy daa. The iegraed IS/GPS deivers he accurae souio copared o he referece. Accordig o he resus, i ca be cocuded ha he IS souio ca be iproved by iigaig he ihere IMU sesor errors wih he aid of iegraig he GPS daa. AKOWEDGEMETS This research is suppored by a SER (aura Scieces ad Egieerig Research ouci of aada) Research Projec. The auhors woud aso ike o appreciae he fu suppor by Prof. Reé Jr. adry ad he coribuios ade by Phiippe avoie ad Erik Poirier i he projec. REFEREES Savage P.G.(), Srapdow Aayics - Par I, Srapdow Associaes, Ic. Savage P.G.(), Srapdow Aayics - Par II, Srapdow Associaes, Ic.

Tiero D. H.(997), Weso, J.., Srapdow Ieria avigaio Techoogy, Ediio II, Isiuio of Eecrica Egieers. hafied A..(997), Fudaea of High Accuracy Ieria avigaio, Progress i Asroauics ad Aeroauics Farre J.(999) The Goba Posiioig Syse ad Ieria avigaio. McGraw-Hi. row R.G., Hwag P.Y..(997) Iroducio o Rado Sigas ad Appied Kaa Fierig, 3rd ediio, Joh Wiey & Sos. Maybeck P.S.(979) Sochasic Modes, Esiaio, ad oro, Voue I. riig K. R.(97) Ieria avigaio Syse Aaysis, Joh Wiey ad Sos. Kourepeis A., oresei J., oey, J., Eio R.(998), Perforace of MEMS ieria sesors, IEEE Posiio, ocaio ad avigaio Syposiu. Kog, X.Y. Deveope of a oiear Psi-age ode for arge isaige errors ad is appicaio i IS aige ad caibraio, Proceedigs of he 999 IEEE Ieraioa oferece o Roboics ad Auoaio, Vo., 999, p. 43-435 Giroux R., adry R.Jr., each., Gourdeau R.,(3) «Vaidaio ad Perforace Evauaio of a Siuik Ieria avigaio Syse Siuaor», The aadia Aeroauics ad Space Joura, Vo. 49, o.4, ISS 8-8, pp.49-6.