Trip Reconstruction Tool for GPS-Based Personal Travel Surveys
Eui-Hwan Chung
Introduction Transportation planning model
Forecast and evaluate transportation scenarios Require good-quality travel survey data
Conventional self-reporting survey method lack of reporting of short trips and actual routes traveled poor data quality of
travel start and end times, total trip times and, location of destination
the amount of detail that it is feasible to ask individuals and households to report is well below that needed for the activity-based micro-simulation models
Introduction Application of GPS for travel surveys
As the GPS receiver is given to respondents, improve the quality of the collected data serve the convenience of both respondents and survey
operators The benefit of GPS [Wolf, 2000]
trip origin, destination, and route data are automatically collected without burden on the respondent
routes are recorded for all trips allowing for the post-processing recovery of unreported or misreported trips
accurate trip start and end times are automatically determined, as well as trip lengths
the GPS data can be used to verify self-reported data.
Introduction Approaches for GPS Application
Electronic Travel Diary (ETD) with GPS For each trip a respondent records the following
information to ETD (Just replace paper) trip mode, vehicle identification, driver identification,
passenger identification, driver and passenger trip purposes, trip start time, finish time (or duration), origin location, destination location,and distance traveled.
In addition to these traditional elements, from GPS Route choice and travel speed can be captured
Introduction Approaches for GPS Application
Passive in-vehicle GPS systems The intent of passive in-vehicle GPS systems
To conduct a passive audit of in-vehicle travel The GPS data will be used in a post-processing step
to the recorded travel diary of the respondent to validate the reported data and/or to determine trip under-reporting rates.
The data can be useful in telephone interview (by refreshing respondent’s memory).
Introduction Approaches for GPS Application
Total replacement of the travel diary with GPS. Use GPS data logger to completely replace, rather than
supplement, traditional travel diaries All essential trip elements are derived through a computerized
process of all GPS data both respondent burden and telephone interview time could be
reduced significantly. Previous research [Wolf, 2000] tried the following
Trip detection (the number of trips) To find Land uses and addresses for trip destinations Trip Purpose Derivation Trip Distance
Purpose Develop algorithms to reconstruct trips of a
traveler holding GPS-logger To automatically identify network links and modes
used
Illustration of Overall Concept
Passenger Car
Walking
Mode Change
Used links
Used Data Sets for Base Map Data sources for map of this thesis
Geographic features of the map DMTI CanMap® Streetfiles Version 6.2
Transportation property of the map 2001 EMME/2 road network V1.0 Transit stop and timetable information from the TTC (Toronto Tra
nsit Commission) Spatial Scope : Downtown Toronto
Used Tool - ArcGIS Version 8.2 Popular and powerful GIS S/W Provides programming interface ArcObject – Component Object Model (COM)
Used Device – GPS(1/2) Global Positioning System(GPS)
Provides 3D coordinates of current position on Earth using artificial satellites
3D coordinates – at least 4 satellites 2D coordinates – at least 3 satellites As distribution of the satellites is wider and the num. of the
satellites that a GPS receiver gets signal from simultaneously is larger, the accuracy of the estimated coordinate improves
Gets speed and azimuth of movement using the Doppler effect
Used Device – GPS(2/2)
GPS used in this research Wearable With a logger – recorder of GPS points which are
collected by a GPS receiver
Identifying Used Road Segment
GPS datapre-processing
Main matchingprocess
Post matchingprocessing
GPSlogger
GISmap
Reformat dataDelete invalid data
Match the GPS point with the link
Arrange the results ofthe matching process
Pre-processing Reformat GPS data
CVS file format (text file) DBF format UTC time Local time Change format of longitude/latitude Add additional fields for the next process
Eliminate invalid data Number of satellites
To get 2D coordinates, a GPS receiver should get signals from at least 3 GPS satellites simultaneously
if #Sat < 3 then delete the row HDOP
Dispersion of satellites from which a GPS receiver receives signal The wider the dispersion of satellites, the better accuracy of the measured
coordinates if HDOP > 5 then delete the row
Main matching process The purpose of this process
Match the GPS point with the link Identify the traveled links based on the respondent’s GPS
data
Find matched a link based on distance and azimuth of moving direction
Road Network
GPS points
Which link should be matched with each GPS point ?
Main matching process Matching Algorithm
Considering both distance between the GPS point and the link and an azimuth of GPS point movements
Use topological information – make use of the geometry of the arcs as well as the connectivity of the arcs
almost perpendicular
Post Matching Process The aim of the process
Make a list of used links
No Link_OID GPS_Start GPS_End FromNodeID ToNodeID MatchType …..0 103 1 4 10 111 102 5 8 10 1323
…..
Link OID: 101
Link OID: 103Link OID: 102
Link OID: 104
01
Start (Point_OID = 1 )
(11)(10)
(12)
(13)
(14)
Identification of Used Modes Available clues for estimation of used modes
List of Used Links Availability of transit Property of used link – e.g.) freeway, one-way road
GIS Map Location of transit stops
Travel Speed From GPS
Limitations of GPS Limitations
Effect of “Cold/warm Start”
Effect of Urban Canyon
Limitations of GPS Limitations
Limitations of GPS Limitations
Difficulty of Getting GPS Signal in a busDifficulty of Getting GPS Signal in a bus
Identification of Used Modes Basically, to estimate modes, good level of quality of GPS data is
required. No GPS Data No Result Good quality of data Elaborate rule
Just depending on GPS Data, it is not easy to estimate all kinds of mode configurations.
Assumptions A trip is one purpose trip. A mode configuration pattern of a trip is one of the following
Walk only, Walk Bicycle Walk, Walk Passenger Car Walk, and Walk Transit Walk.
Both trip ends are not in urban canyon area. There is no cold start of GPS receiver.
Identification of Used Modes Process for mode identification
Two important clues Location of mode change points (ending point of the first walking, and the starting
point of the last walking) Maximum speed from GPS data
ProcessStep 1) Find two mode change points, the ending point of the first walking and
the starting point of the last walking segments. Step 2) If two mode change points exist, go to step 3. Otherwise, go to step 4Step 3) If both points are in a buffer area of bus stops, the used mode is a bus.
The size of the buffer is set at a radius of 40m from a stop.Step 4) If maximum speed is faster than 32 km/h, the used mode is a
passenger car.Step 5) If maximum speed is faster than 10 km/h, the used mode is a bicycle.
Otherwise, the used mode is a walk
Evaluation Test suggested methodology using real GPS data Made 60 trips with GPS receiver
Reproduce the 60 trips based on existing TTS data Randomly sample each trip from the data O/D and Mode Route of a trip was decided by a respondent
Both O and D are in downtown Toronto area To satisfy assumptions Mode Configuration
10 Bus 24 Passenger Car 23 Walk 3 Bicycle
Evaluation
Identification of used links % of correctly identified links: 78.5 % % of un-detected links: 21.5 % % of incorrectly identified links: 0 %
Identification of used modes % of correctly identified modes: 91.7(55/60)