Developing a Mobile Transportation Survey System

1
Developing a Mobile Transportation Survey System

• Ian Bates
• Automatica Ltd

2
History and Introduction

• c.1991 Manchester Study (paper based).
• Early 2001 study proposed by Tyne and Wear Joint
  Transport Working Group.
• Mid 2001 Automatica Ltd commissioned to develop
  system.
• Feb 2002 HP Jornada based Alpha software released
  and tested.
• Sept 2002 HP Jornada based Beta version field
  tested with 15 interviews.

3
History and Introduction (2)

• Late 2002 the Jornada withdrawn and Itronix
  Go-Book selected as replacement.
• Nov 2002 Extended Pilot - 30 Interviews.
• March 2003 Training full study began 10-15 SMS
  Ltd Interviewers in field.
• March 2005 around 5000 interviews completed.

4
Designing the System

• Survey Design
• Random sampling of entire population
• Panel survey to analyse change.
• Questionnaire Design
• Complex questionnaire
• Household, Person, Trip components
• 30-45 min interview of entire family
• Address capture for trip origin/destination
• Public transport route information.
• Cost of interviewing 100/Interview!

5
Designing the SystemChoosing a Device

• Full keyboard required for postal address entry.
• Relatively large screen area for complex
  questionnaire.
• Features of the HP Jornada 720
• It has a useful 640x320 half VGA form factor,
  giving it a relatively large screen area for a
  handheld.
• It has a stylus type touch screen.
• It is small and light and can be easily carried
  by interviewers. Also at that time it was
  slightly cheaper than a laptop.
• The device has a built in modem to enable the
  uploading of data.
• It has a built in full QWERTY keyboard that
  despite being a little small enables the quick
  entry of textual information.
• Microsoft had recently released a free suite of
  development tools for this type of device. This
  included a Visual Basic and Visual C compiler
  environment.

6
The Chosen Mobile Devices
HP Jornada 720
Itronix Go-Book
7
Designing the SystemData Storage

• Data stored on central SQL database.
• Client server system used
• Server uses Microsoft SQL Server 2000
• Mobile Devices running MS SQL Server CE.
• Communication using modem via internet based
  protocol (Remote Data Access).
• Mobile device uses 128 MB Flash Cards.

8
Household Survey Database
Household Survey
Person Survey
Basic Household Information i.e. Cars/Bicycles,
Members present, Property Type
Trip Survey
Person Information i.e. Age, Gender, Work,
Disabilities, Car Licence
Trip Stage Information
Trip Information i.e. Start End Address, Start
End Time, Purpose
Trip Stage Information i.e. Modes of transport,
bus/metro/rail routes, cost, car parking
information
9
Designing the SystemQuestionnaire Software

• Colour coded tree like navigation screen for
  scheduling, monitoring and controlling the
  interview process.
• Automatic, internet based synchronisation
• Transfers complete interviews to server
• Updates device with new sample as allocated.
• Written using Embedded Visual Basic.

10
Device Tree Display
11
Designing the SystemTrip Interview Software

• Recording trip information
• Captures address information for trips made by
  people in the household
• Trip purpose i.e. going shopping or escorting
  children to school
• Multiple mode multiple stage trips i.e. Car, Bus,
  Walk.
• Records public transport details of trip stages
  i.e. fare, ticket type, waiting time, route.

12
Trip Information Display
13
Designing the SystemAddress Finder Software

• Address finder allows coding of trip address at
  time of interview.
• Uses a database based on Post office Address File
  (PAF) for Sunderland, Durham, Newcastle postal
  areas.
• Organisation / Building name searching.
• Street name searching to obtain postcode.

14
Searching for an Organisation Name
15
Designing the SystemAddress Finder
Technical/Limitations

• 20MB Organisation file is stored on flash card.
  Uses MS SQL Server CE 2.0 engine.
• Searching may be slow if first character of
  organisation is not known i.e. searching for all
  schools in a town using a wild card.
• Visual Basic software API does not allow a long
  running query to be cancelled this can hang
  device.

16
Trip Stage Information
17
Public Transport Information
18
Designing the SystemPublic Transport Route Finder

• Public transport route finder allows bus, metro,
  rail and ferry routes and related stop/station
  information in Tyne and Wear to be located.
• Can search for all routes from a stop/station
  name.
• Uses route data from the Nexus PAJIN database
  system.

19
Searching for a Bus Route
20
Designing the SystemPublic Transport Route
Finder Notes

• When searching for routes from a main bus
  station, search times may be long as many routes
  are produced.
• Most effective is to enter bus route numbers, if
  known by interviewee.

21
Designing the SystemSpecial Features

• Special Features added as a result of
  Interviewers requests in pilot phases
• Copy trip facility allows copying of trips
  shared between household members i.e. several
  children going to school with parent.
• Pull down lists of previously entered addresses
  for a household are available to save re-entering
  addresses.

22
Uploading and Downloading Information
23
Analysing Data

• Data is immediately available on central server
  on upload from devices.
• Custom Web Site for viewing/analysing
• Survey Progress Statistics
• Interviewer Impact using Multiple Classification
  Analysis
• Various types of Trip Analysis.

24
(No Transcript)
25
(No Transcript)
26
ConclusionsBenefits of the System

• Survey contractor - SMS Ltd (Survey Marketing
  Services)
• Interviewers find the system easy to use.
• Survey administration, including sample
  allocation, is simplified compared to paper based
  surveys.
• Interviewer progress and performance may be
  monitored via the web site.
• Client - Tyne and Wear Joint Transport Working
  Group
• The data collected may be analysed and queried
  immediately, either via the survey web-site, or
  intranet at Newcastle City Council.
• Data is securely held on the server environment,
  so in the event of mobile device problems
  occurring possible data loss is limited to a
  maximum of a single days work held on the mobile
  device.

27
ConclusionsLessons Learnt

• Mobile devices have short lifetimes and this can
  cause problems when sourcing devices and
  replacements.
• Microsofts Embedded Visual Basic language used
  to develop the software does not have adequate
  internal checking. This makes the development of
  a complex piece of software more difficult.
• Battery life is a major issue as the charge in
  mobile device batteries rarely lasts the time
  claimed in the manufacturers specifications.

28
Future Work

• Interviewing using system is ongoing for next few
  years
• Comparison of data by area and phase
• Statistical normalisation of data to eliminate
  interviewer bias.
• Development of website to provide Tyne and Wear
  partners with key indicators.

29
Thank you for Listening!

• Ian Bates ian_at_automatica.co.uk