A Portable and Cost-effective Configuration of Strap-down INS/GPS for General-purpose Use

1
A Portable and Cost-effective Configurationof
Strap-down INS/GPSfor General-purpose Use

• Masaru Naruoka (Univ. of Tokyo)
• Takeshi Tsuchiya (Univ. of Tokyo)

2
Outline

1. Background we need precise, small, light and
   inexpensive navigation system.
2. Method my system
3. Evaluation how precise?
4. Conclusion

3
Background (1/5)Many Needs for Precise
Navigation data

• Many applications for requiring precise
  navigation data (position, velocity, attitude)
• Of course, navigate aircrafts, spacecrafts
• Observe cars, trains, etc.
• Control robots, UAVs

Can we apply accumulated navigation technologies
of aircrafts to these applications?
4
Background (2/5)One of technologies INS/GPS

• INS/GPS Navigation System

Inertial Navigation System (INS)
Global Positioning System (GPS)

High update ratio but Accumulate Error
Cancel Error but Low update ratio
Cancel Error and High update ratio
5
Background (3/5)Mechanism of INS/GPS
the laws of motion
triangle surveying
INS
GPS
Inertial Sensors
Receiver
Position Velocity Attitude
Acceleration Angular Speed
Satellites
Position Velocity
Radio wave
Integration
Position, Velocity, Attitude
INS/GPS
6
Background (4/5)Traditional vs. Developing
INS/GPS

• Traditional

• Now developed

Precise?
Ultra Precise (Error lt1m, lt1deg)
Trade-Off
motivation
Big (gt 1000 cm3) Heavy (gt 1 kg)
Small (lt 1000 cm3) Light (lt 1 kg)
Only Aircrafts and Spacecrafts!
Cost-Effective (lt 100K)
Expensive (gt 100K)
7
Background (5/5)Goal of my study

• Meaningful to discuss about the trade-off between
  precision and other specifications.
• The goal of my study
• Develop as small, light, cost-effective INS/GPS
  system as possible.
• Investigate its precision correctly

8
Method (1/7)Components
Do not use
Use

• Big, heavy, expensive dedicated components
• Ring laser gyro
• Military-use GPS

• Small, light, inexpensive components
• MEMS inertial sensors
• Civil-use GPS

9
Method (2/7)MEMS Sensors and Civil-use GPS

• MEMS inertial sensors
• Electronic circuit and sensing element integrated
• Small(1 cm2), Light(lt1 g), Inexpensive(lt100)
• BUT, an INS device using MEMS inertial sensors
  accumulates error very quickly.
• Civil-use GPS receiver
• Mainly for car navigation system
• Small(10 cm2), Light(lt10g), Inexpensive(100)
• AND good precision (Error 1020m)

10
Method (3/7)INS/GPS Algorithm

• Strap-down configuration
• No need for any mechanical gimbals
• Integration by extended Kalman filtering (EKF)
• Loose-coupling require small calculation power
• Use quaternions actively
• Mathematically simple model to compensate for
  large MEMS sensor error
• Eliminate singular points derived from Euler
  angles

11
Method (4/7)Equations equations of motion for
INS

• Velocity (3North, East, Down Speed States)

Acceleration
Gravity

• Position (4Latitude, Longitude, Azimuth
  1Height 5 States)

quaternion

• Attitude (4Roll, Pitch, Heading States)

Angular Speed
12
Method (5/7)Equations Linearization for EKF
Obtain linearized form for EKF by following
substitution to the equations of motion
quaternion
Quaternion linearization with keeping the norm
unity

• Jacobian i.e. Additive (4 States)

• Multiplicative (3 States)

13
Method (6/7)Equations EKF
When INS update
When GPS data is obtained

• EKF Time Update

• EKF Correct

quaternion
14
Method (7/7)Overall View
Strap-down configuration
MEMS sensors
Leverage quaternion for system modeling
15
Evaluation (1/11)Outline

• Prototyping
• Based on my proposed system
• Calibration is performed
• Test for Precision
• Compare the prototype with an existent precise
  navigation device

16
Evaluation (2/11)Developed Prototype
Size 100 cm3 Weight 30 g Cost 300
(w/o structural element)
The prototype shows my system is small, light and
low-cost
17
Evaluation (3/11)Detail of Prototype
18
Evaluation (4/11)Calibration for MEMS INS

• Temperature Drift

• Misalignment

rotating
settling
main error source of MEMS INS that can be easily
removed
19
Evaluation (5/11)Calibration of Prototype

• The result of calibrations

• Temperature drift

• Misalignment

slope not 0 !
slope not 0 !
Y,Z
X
Temperature
True Angular Speed (X-axis)
vs.
vs.
Sensed Angular Speed (X, Y, Z-axis Gyro)
Sensed Acceleration (X-axis Accelerometer)
20
Evaluation (6/11)Test for Precision

• Comparison of the prototype with GAIA (2006/06)
• GAIA an ultra high-precision INS/GPS device
  developed by Japan Aerospace Exploration Agency
  (JAXA)
• Error lt 1m in absolute position
• In flight of an experimental aircraft, MuPAL-a of
  JAXA

21
Evaluation (7/11)Scene of Test
GAIA
MuPAL-a
Prototype
22
Evaluation (8/11)Results of Test

• Results (The Prototype Red, GAIA Green)

Position (3D)
Velocity
Attitude
Nearly equal to GAIA
23
Evaluation (9/11)Detail of Results
Statistical Summary of the error of the prototype
by reference to GAIA
Position
lt 10m
Velocity
lt 2 deg
Attitude
gt 10 deg
24
Evaluation (10/11)Summary and Discussion of Test

• Error lt10m(Position), lt2deg(Roll, Pitch)
• Precise enough for general-purpose use
• Heading is the worst (Error gt10 deg)
• Effect of frequency mode of dynamics
• Roll and Pitch is comparatively high mode (gt 1Hz)
• Heading is low mode (lt 1 Hz)
• Overlap with low mode noise that cannot remove
  easily (for example, zero-point change)

25
Evaluation (11/11)Effectiveness of Calibration

• With calibration

• Without calibration

Example Roll History
Calibration works well.
26
Conclusion

• My Navigation System
• small, light and cost-effective INS/GPS system
  for general-purpose use
• Strap-down configuration using MEMS sensors and a
  civil-use GPS receiver
• Temperature drift and misalignment calibrated
• Use EKF and Quaternion
• The test shows it is precise enough for
  general-purpose use Error is under 10 m in
  position, and 2 deg in roll and pitch.

27
Future Work

• Fight against low frequency noise
• Time-Frequency Analysis
• Wavelet multi resolution analysis
• Wavelet de-nosing
• Other compensation system
• Earth magnetism sensor for attitude etc.