the geta sandals: a footprint location tracking system

The GETA Sandals: A Footprint Location Tracking System

Kenji Okuda, Shun-yuan Yeh, Chon-in Wu, Keng-hao Chang, and Hao-hua Chu

National Taiwan University

What is GETA?

a kind of the traditional Japanese sandals

Outline

Motivation Related works Basic idea Design, implementation and evaluation

(three versions try and error)

Conclusion Current works

Motivation Infrastructure cost (barriers for the user) (deployment, calibration, maintain)

Active Badge: IR receivers Active BAT: Ultrasonic receivers + RF Cricket: Ultrasonic beacons + RF RADAR: WiFi network Smart floor: pressure sensors

Goal Infrastructure free (not succeed so far in this paper-> low infrastructure) High accuracy

Related Works

Lee et al. proposed a method by recognizing a sequence of incremental

motions. E.g. 5 steps north followed by 14 steps east.

Only can tell from place to place e.g. living room, bedroom

Point research provides a vehicle self-tracking system

Basic idea of Foot-print Approach

( 0, 0 )

( 180, 200 )

Sp

S1

S2

S3

S4

S5

Sp=S1+S2+S3+S4+S5

How to sum up those vectors mathematically

Coordinate system 1

Coordinate system 1’

(x1, y1)

(x2, y2)

(xc2, yc2)

(xc1, yc1)

θCoordinate system 2

Need coordinate transformation for each left foot step.

Three try-and-error versions

Design version I

Bottom view side view

Ultrasonic Transmitte

rs

Ultrasonic

Receivers

Why using two transmitters

P1 (known)

P2 (Unknown)

T1d1

d2

d3

d4

Can be rotated!!

fixed

(fixed)

Why using two transmitters

P1 (known)

P2 (determined)

T1

Can Not be rotated any

more !

T2

Problems of Design I

Poor accuracy! The interference of the signals from two

transmitters. Measures the incorrect vectors Miss-detection of the user’s steps All the calculations become failure.

Can not distinguish the user is moving forward or backward!

Design version II

Bottom view side view

Ultrasonic Transmitte

r

Ultrasonic

Receivers

Pressure sensor

Orientation Sensor

Design II Performance Evaluation

0

2

4

6

8

10

0 100 200 300 400 500 600

distance (cm)

erro

r %

0.43(steps/sec)

0.7(steps/sec)

1.3(steps/sec)

0

10

20

30

40

50

0 100 200 300 400 500 600

distance(cm)

erro

r d

ista

nce

(cm

)

0.43(steps/sec)

0.7(steps/sec)

1.3(steps/sec)

Two Main Error Sources

θerrP1 P2

Real path Calculated path Error Displacement

P2

Steps

Error

Real path Calculated path Error Displacement

P1

Fig1. Ultrasonic device err Fig2. Orientation sensor err

The solution of design II

Passive RFID tag

Adds a RFID reader in the GETA and put some tags in the environment.

Real Path of the user Calculated Path of the user

Design version III Orientation

SensorRFID Reader

Pressure Sensor

Ultrasonic Receivers board (with 2 Ultrasonic receivers)

Ultrasonic transmitter

Hardware Sensors of Design III

Pressure sensors Phidgets

Ultrasonic device NAVINote (an electronic pen product) Resolution : 0.2 mm

Orientation sensor InterSense InterTrax2 Resolution : 0.02 degree

RFID Reader SkyeTek M1 Read range: ~5cm

Design III Performance Evaluation

Doesn't step over thetag

Doesn't step over thetag

020406080

100120140

0 500 1000 1500 2000 2500 3000

distance (cm)

erro

r dis

tanc

e (c

m) one tag per 3 m

one tage per 5 m

The positioning error under different tag density over the walking distance.

Conclusion

A interesting self tracking method. Low infrastructure cost. The error of the orientation dominants our

system accuracy. Still have some limitations

E.g. climbing the stairs, walking crossover..

Current Works

Increase the accuracy

(reducing the orientation error) Solve the obstruction problem Going down/up stair problem Enhance the wearability (wireless)

Thank you

Questions?

or send me E-mail r93124@csie.ntu.edu.tw

Thank you !