1 Use of GPS in Travel Surveys Cost 355, WG3 – Annecy Conference, 26 May 2008 Philippe Marchal and Pierre-Olivier Flavigny (INRETS) Shuning Yuan (INRETS.

1

Use of GPS in Travel Surveys

Cost 355, WG3 – Annecy Conference, 26 May 2008Philippe Marchal and Pierre-Olivier Flavigny (INRETS)

Shuning Yuan (INRETS and ISL)

2

Cost 355 contributions

Combining person based GPS tracking and prompted recall interviews for a comprehensive investigation of travel behaviour adaptation processes during life course transitions (Michael Flamm, Christophe Jemelin & Vincent Kaufmann– EPFL - LASUR Lausanne)

Development of an integrated person based GPS-aided travel survey system: solutions with and without GIS data layers (Martin Lee-Gosselin, Université Laval, Québec & The PROCESSUS Network, Canada)

The project « Lille Experiment » by ISL-Lavialle, Survey Institute (Shuning Yuan, INRETS and ISL)

GPS Subset in the French National Travel Survey (ENTD 2007-2008)

3

Travel behaviour adaptation processes during life course transitions: aims

Provide a better understanding of the processes underlying behavioural changes during life course transitions

Reflect on possible modes of intervention for inducing behaviour change during life course transitions

Test an innovative survey design, combining person-based GPS tracking and qualitative prompted recall interviews

4

Travel behaviour adaptation processes during life course transitions: first insights

Survey design offers unprecedented research opportunities for analysing learning processes (destination choices, route choices, experiences of new travel modes, etc.)

Social networks and personal practice strongly influence destination choice in unknown environment; some people use internet for identifying new activity locations

However: in general, people tend to minimize cognitive effort with regard to reorganizing their travel behaviour (except people who are concerned about environmental problems)

5

The project « Lille Experiment » by ISL-Lavialle, Survey Institute

6

Introduction of the Lille project

• Establish a comparison between the data collected in the

study « Mobimétrie 2004 » and the traces collected by

GPS in this project. • The experiment in Lille and its surrounding area in 2006 1 second/point

• Two waves (57+145 interviewees)

• Eurisko GPS receiverL1, C/A codes, 8 channels, 24 hours Autonomy , 16M SD card

five buttons (Walk or bicycle, automobile, public transport, study or work, home), no DOPs recorded

7

Method for filtering the invalid recordings

The Method Acceleration-Speed (MAV)The Method Acceleration-Speed (MAV)Principal ideas

If “Vb” >> “Va” Then “C” is badly positioned.

8

Method for filtering the invalid recordings

9

Results of the project

Number of respondents

Number of

weeksNumber of recordings

Number of traces

obtained

Number of places found

Wave 1 57 114 7742534 3487 4423

Wave 2 145 187 13012716 5593 6864

Total 202 302 20 755 250 9080 11287

NumAdr NbRecordNbRecordFiltre % valid

21571 6520 5091 78

20027 17305 14359 82

21134 31467 27526 87

17499 162015 139525 86

15796 276904 250826 90

17992 324216 301363 92

19924 618692 567716 91

10

Results of the projects

  Number of respondents

Total traces

Average Number of

traces /day/capita

Average Distance

(km)

Average speed(km/h)

Survey 1week

83 2 360 4.1 5.3 15

Survey 2weeks

76 4 079 3.8 6.9 18

Total 159 6 439 3.9 6.3 17

11

Distribution of the traces

on the main modes Number

of traces

Traces

(%)

Average Distance

(km)

Average Duration

(min)

Average Speed (km/h)

Auto, moto 2.103 36% 7,0 17 21

On foot, bike, roller

1.433 24% 5,6 17 15

Public Transport

131 2% 7,6 24 13

Undefined2.248 38% 5,5 18 14

Total/

average5.915 100% 6,1 17 17

12

Purpose of the traces

Nomber of traces

Traces

(%)

Average Distance

(km)

Average Duratio

n

(min)

Average Speed

(km/h)

Work-Study

1.594 27% 5,9 18 17

Others 3.731 63% 6,6 18 17

Undefined 605 10% 2,7 13 10

Total/

Average5.930 100% 6,0 17 17

13

Results of the project

Effect of survey duration on collected traces numberOverall : 6439 traces

389366

317

272296

374346

261 247

291323

294

340308

366

291 307245

251282273

0

50

100

150

200

250

300

350

400

450

14

Overview of the French National Travel Survey (ENTD 2007-2008)

Sample ofdwellings

Respondents households

Individuals living in the selected household

Vehicle fleet[car, motorbike, moped, bicycle]

Kish

Daily trips

Long distance trips

GPS component

Vehicle diary

Biography

Primary utility of travel

Design

Key figures

Sample size : approx. 19 000 responding households

Duration : one full year (6 waves)

2 visits

Total length of interview : 115 mn

15

Firstface-to-face

interview

Secondface-to-face

interview

The interviewer givesthe « GPS Pack » to the

respondent and explains how touse it

Global survey

Volunteer for GPS ?(and GPS unit available)

The respondent travels and the GPSunit records trips

The respondent gives the« GPS Pack »

back to the interviewer

At the end of the global CAPI,brief additional questionnaire

CAPI-GPS

The interviewer checks the GPS,reloads it: the equipment is ready for

a new interview

Immediately : the interviewerdownloads the GPS data

on his computer(Bluetooth connection)

No

General scheme of the GPS component

GPS component

Day D Day D + 7 (or more)

Yes

16

Equipment and documents for respondents and interviewers

GPS unit

A/C adapter

“User manual” (one page)

Given to the respondents Given to the interviewers

SD Card MemoryPlaced in the laptop card reader from the beginning to the end of the survey

Bluetooth USB AdapterInserted before downloading the data (at each interview)

Manual for the Interviewer (2 pages)

17

17

Presentation of the GPS units selected

• Passive monitoring: the respondant has no graphical interface, for road safety and to avoid an influence of GPS on travel behavior, etc.

• Only one button: on/off. The respondent has the possibility to skip some trips, if desired (for confidentiality reasons)

• No data transmitted on real-time: the device is only a datalogger. Data transmitted to the interviewer's laptop during the second visit, and deleted inside the unit

• Two types of GPS unit: 100 « normal » and 70 modified, with movement detection and blinking mode

18

French Regions with aGPS sub-sample

19

CAPI-GPS design

Computational analysis ofdownloaded GPS data

Selection of days and trips

Day without GPS record

“During this day, we have nodata recorded, could you tellwhy ?”:

-I’ve forgotten the GPS unit-Unit off-Absolutely no trip-Problem with the GPS unit(battery)

Recorded trips duringone selected day

Prompted recall with tripcharacteristics (day of the week,date, departure time, duration,nearest origin and destinationpoints, time spent), then thefollowing questions:

1 : « For what purpose ? »2 : « Number of accompanyingpersons ? »3 : « Means of transport used ? »

20

A data collection framework with three combined measurement tools

How trips caracteristicsfor days with no CAPI-GPS and limited conventional questionnaire

can be derived from raw GPS data ?

Conventional questionnaire GPS traces CAPI-GPS

- Socio-economic caracteristics

- Last w-e trips

- Last day trips

- etc.

Time-stamped positions Days without any GPS record:Why ?

Selection of recorded trips:-Purpose ?-Number of accompanying persons ?-Transport modes ?

How trips caracteristicsfor days with no CAPI-GPS

can be derived from raw GPS data ?

21

Work in progress

•Trip estimation from the GPS recordingsoTraces segmentation (smaller and homogenous segments)

oAnalyse segment ends, and link them to a trip

•Modes and trip purpose estimation

•Incomplete and missing trips estimation

•Modal map-matching

22

Thank you !

INRETS French NTS Team:

Jimmy ARMOOGUM,Jean-Paul HUBERT,Philippe MARCHAL,Pierre-Olivier FLAVIGNY,Shuning YUAN,Sophie ROUX,Jean-Loup MADRE,Francis PAPON