GPSCell Phones For Trip Data Collection

1
GPS-Cell Phones For Trip Data Collection

• A test of out-of-the-box technology
• Results and limitations
• A test of enhancements made to improve quality of
  data
• Additional suggested enhancements
• Is this technology ready for current use?

2
Accutrack Enhanced Technology Test

• Use of Accutrack
• All tests with small (and friendly) sample
• New York, Boston, Rural Vermont New Hampshire
• Tests of Drive, Walk, Bike, Rail
• Enhancements Tested
• Phone-Side Software
• Server-Side Software

3
Out-Of-Box Technology - Accutrack

• General Requirement
• Motorola i730 iDEN handset
• Nextel unlimited data access plan
• Accutrack software installed on iDEN phone
• Link to Accutrack server
• View on website for realtime
• Download data for further analysis

4
Bike Trip Results

• Commonwealth Avenue in South End of Boston
• One minute drop correctly represented captured
  bike trip route

5
Walk Trip Results

• Boston Back Bay Station to a major downtown
  office building
• Some problems with Urban Canyons
• Locations not accurate to roadway
• Inferentially accurate

6
Car Trip Results

• Driving from White River Junction, Vermont to
  Boston, MA
• Cell coverage is spotty
• One minute was sufficient for highway trips
• Require greater precision for non-highway trips

7
RSG Enhancement Cell Phone Software

• Data Storage and upload
• cache data on phone and upload when signal is
  strong
• Compress data on phone side to limit upload (same
  location).
• Dynamically alter location upload times based on
  speed distance
• Assisted GPS for enhanced location information

8
Nightly Data Validation

• Recall using web-based tool
• Correction of routes
• Identification of stops
• Addition of meta-data
• Trip type
• Auto occupancy
• Other recall features
• Update respondent regarding phone battery
• Show history and support association to previous
  trips

9
Meta-Data Enhancements

• Collection methods
• Voice activated trip purpose information
• Voice prompting is this a stop
• Algorithmic post-processing
• Heuristics for linking O-D where either urban
  canyons or transit trips cause loss of data
• Probabilistic (Bayesian) models of activity and
  location based on web information of surrounding
  options