GPS/INS Computing System

Company

LOGO

Mid semester presentationSpring 2008/9

Performed by:Alexander Pavlov David Domb

Supervisor:Mony Orbach

GPS/INS Computing

System

Agenda

1. General overview

2. Our Project

4. What’s Next…

GPS/INS Computing System

3. The Design


General overview

“Even Noah got no salary for the first six months partly on account of the weather and partly because he was learning navigation.”

Mark Twain

Theoretical Navigation Algorithm

0 •Initialization1 •Particle Propagation2 •Particle Update & Normalization3 •State Estimation4 •Effective N calculation5 •D computation6 •Re-sampling7 •Regularization8 •Weight Re-computation


Developed in the “Technion” and Implements the tightly coupled INS/GPS navigation unit, with the particle filter.

The algorithm stages:

Project Goals

Establishing the efficiency of the particle filter based, tightly coupled INS/GPS navigation unit realization.

Designing an efficient real-time particle filter based, tightly coupled INS/GPS navigation unit.


GPS Computing System

Our Project

General• Project will be performed in 2

stages. First part in this semester.• Project will be performed by

several work groups• Our group will implement Particle

Propagation and State Estimation stages in this first part.

• Both stages need to be performed whit in 0.01 sec, regardless of other stages performance.


Group Project Goals – Part 1Learning GPS/INS navigation using Particle Filter algorithm

Learning VHDL language

Learning FPGA environment

Implementation of Particle Propagation and State Estimation stages of algorithm



TheDesign

Design guidelinesConstrains: large amount of calculations Limited hardware real-time results

Selected solution:Combining Parallel processing whit

Pipelines.


X data structureTrue State Output Record

Field Sign bit Number bits Fraction bitsPosition 0 0 28

0 0 28dummy 0 16 0

1 14 9Speed 1 13 10

1 13 101 13 10

Quaternion 1 1 221 1 221 1 221 1 22

Acceleration offset

1 0 231 0 231 0 23

Dreidel offset 1 0 231 0 231 0 23

GPS Receiver offset

1 14 90 24 0

Dummy 0 24 0


h

NV

EV

DV

1q

2q

3q

4q

X

Y

Z

X

Y

Z

bb

INS data structureTrue State Output Record

Field Sign bit Number bits Fraction bitsAcceleration 1 5 42

1 5 421 5 42

Angular rates 1 2 451 2 451 2 45


xV

yV

zV

x

y

z

W data structureTrue State Output Record

Field Sign bit Number bits Fraction bitsWeight 0 0 24Dummy 0 24 0

0 24 00 24 0

w

Solution – Top design


Weight vector

Particles propagation

unit

State estimation

unit

Estimated State Vector

[1..18]

xN Extended State Vector

[1..18]

Extended State Vector

[1..18]


[1..18]

Controller

Controller Algorithm


While “FIFO” is NOT empty:• Every 5 clock cycles, send a new particle

from the “FIFO”, into the “TOP_6_PROP” (asserting the “START” signal to ‘1’).

Keep count of “START” signals given.Keep count of “FINISH” signals from the

“TOP_6_PROP”.For every “FINISH” signal, send the

matching weight vector and new propagated particle to the “TOP_ESTIMATION”.



Weight vector


unit

State estimation

unit


[1..18]


[1..18]


[1..18]


[1..18]

Controller

Particle propagation unit


ParticlePropagation

UnitX(0:439)

INS(0:287)

X_out(0:439)

clockreset

start

finish



PropagationUnit

1

PropagationUnit

2

PropagationUnit

6

6 particles to 1

MUX

Propagationtimingcontrol

Propagation timing control


Every “START” = ‘1’ : counter = counter +1

Propagation unit i starts when: “START” = ‘1’

ANDcounter mod 6 = i.

“FINISH” = ‘1’ when:“finish_i” = ‘1’ for all i.



PropagationUnit

1

PropagationUnit

2

PropagationUnit

6

6 particles to 1

MUX

Propagationtimingcontrol

Single particle propagation data flow

Format inputs to 48 bits

Calculate trigonometric

functions• Latitude sin/cos

Format trigonometric

function output to 48 bits

R_E, R_e, R_N calculation

Denominator calculation

• d_longitude denominator• d_latitude denominator

Dividers• d_longitude• d_lattitude• R_e

ParticlePropagation


Propagationflow

control

Propagation flow control




Weight vector


unit

State estimation

unit


[1..18]


[1..18]


[1..18]


[1..18]

Controller

Estimation unit


EstimationUnitX(0:439)

w(0:23)

Estimated_DATA(0:439)

clockreset

New Data In

Estimation Ready

Estimation unit


W*XW

X

ADDER

Timing Analysis


1 particle LATENCY – 50 clock cycles (from “start” to “finish”) of propagation and weighting (according to simulation).

Propagation stage LATENCY – 45 clocks. Estimation stage LATENCY – 5 clocks.

With a pipeline (Throughput) of 5 clocks, and 6 parallel propagation units : 30,000 particles in 105,050 clocks == 7.5 ms @ 20Mhz.

COMMENTS


NO sin/cos blocks:the design uses a “DUMMY” block with a latency of 30 clocks and no throughput.

The estimation of the quaternion matrix is left to be resolved by another grope (by software). The matrix is part of the design’s output.

MID-Results


According to the initial timing analysis, we will probably be able

to meet the timing demands - “with time to spare”.


What’sNext…

Things to do


Synthesis.Simulations and testing on the board.Final report.


GANTT

GPS/INS Computing System

Documents

Transcript of GPS/INS Computing System