Traffic Flows from Airborne Platforms

1
Traffic Flows from Airborne Platforms
Mark Hickman University of Arizona International
Workshop on Satellite Based Traffic
Measurement Berlin September 9, 2002
2
Outline

• How has airborne imagery been used for traffic
  monitoring and management?
• What research is going on now?
• What does the future hold?

3
History in the US

• 1927 Aerial photography used to measure highway
  congestion (traffic density)
• 1940s 70s Traffic research from airborne
  imagery
• Variety of data collection methods
• Applications areas traffic flow theory, platoon
  behavior, intersection operations, accident
  analysis, parking studies, O-D flow estimates,
  network performance assessment
• Manual data reduction and analysis

4
History in the US

• 1990s Limited work
• Fixed-wing aircraft for level-of-service analysis
• Experiments in intelligent transportation systems
  (ITS) applications for airborne traffic
  management
• 2000s National Consortium on Remote Sensing in
  Transportation - Flows

5
National Consortium on Remote Sensing in
Transportation - Flows

• Rationale for research
• Current sensors and image processing technology
• Mobility and non-intrusiveness of sensors
• Quality and quantity of data
• Goal Improve efficiency of transportation
  system planning and management by integrating
  remotely sensed data with ground-collected data

6
Research Themes of NCRST-F

• Traffic Monitoring
• Traffic Management
• Freight and Intermodal Analysis
• Common Methodological Issues

7
Traffic Monitoring and Management

• Use remotely sensed data on-line in real time /
  near real time to reduce traffic congestion
• Use remotely sensed data off-line for performance
  monitoring and to develop strategies in response
  to recurring traffic congestion

8
Framework
Sensor
Traffic flows
Use video and camera images, image processing and
algorithms to measure traffic variables Major
characteristic Develop new techniques based on
spatial characteristics of traffic flows
9
Focus of Research at University of Arizona

• Develop off-line techniques
• Measures of flow, speed, density, intersection
  delay
• Methods to determine highway level of service
• Information from vehicle trajectories
• Visualization of traffic from airborne imagery
• Develop on-line techniques
• Individual vehicle speeds and trajectories

10
Data Collection Setup
11
Applications for Highway Level of Service

• Methodological issues
• Data collection procedures
• Manual image processing techniques
• Traffic variable extraction and analysis
• Traffic variables of interest
• Freeway densities
• Arterial travel times
• Intersection delay

12
Freeway Experiment

• 10 km along I-10
• Morning peak
• Helicopter, video camera, GPS
• Aircraft speed 100 km/h
• Aircraft height 300 m
• Field of view 275 m

Source Yahoo! Maps
13
Freeway Video
FREEWAY.WMV
14
Analysis of Freeways

• Level of service measure Density
• Passenger cars per unit distance per lane
• Proposed method measures LOS directly
• Identify freeway segment types
• Determine number of lanes, length of roadway
• Determine number and mix of vehicles from imagery
• Compute density and LOS directly

15
Freeway LOS Results
16
Signalized Intersection Experiment

• Site Speedway Boulevard and Euclid Avenue
• 3 min study period (2 cycles)
• Time 815 a.m.
• Field of View 400 m
• 10 sec vehicle counting interval

17
Intersection Video
INTERSECTION.WMV
18
Analysis of Intersections

• Field data procedure for stopped delay and
  control delay taken from the Institute of
  Transportation Engineers (ITE)
• Hovering and/or fixed-wing aircraft

19
Intersection LOS Results
20
Urban Arterial Experiment

• Site Speedway Boulevard
• Time Shoulder of peak (830-900 a.m.)
• Field of view 250 m
• Simultaneous ground travel time data collection
  test car and video cameras at end points

21
Arterial Street Methodology

• Data collection
• Calculate mean travel time for each run
• Compute average (space-mean) speed for level of
  service

22
Arterial Video
ARTERIAL.AVI
23
Arterial Methodology

• Produces higher number of observations than test
  car
• Eliminates driver subjectivity
• Captures within-platoon, between-platoon
  variability

24
Arterial Travel Time Results
25
Other Applications

• Additional traffic information from the aerial
  video
• Turning volumes
• Lane utilization
• Vehicle spacing
• Vehicle trajectories
• Incident effects
• and queuing

26
(No Transcript)
27
Incident Management
28
Technology Development

• Improved imagery characteristics
• GPS used to geo-reference data
• Inertial measurement systems capture camera
  position
• Real-time image transmission to ground
• Automated image processing

29
Current Technology Research

• Orthographic data with GPS, IMU
• Unpiloted aircraft

http//www.geodatasystems.com/
30
Automated Image ProcessingReal-time Speed
Determination

• Use image properties to identify vehicles
• Register sequence of images
• Manual determination of ground control points
• Automated image registration using GPS and IMU
  data
• Subtract images
• Match vehicles and estimate speeds

31
Image Processing Approach
32
Example
2 sec sample frames
33
Example
Georeferencing
Vehicle Matching Individual Vehicle Speeds
34
Automating Image ProcessingRegistration of Video
Imagery

• Idea eliminate effect of video camera movement
• Method automatically register frames using fixed
  locations in the image
• Result smooth point of view in imagery, and
  capability of individual vehicle tracking

35
Vehicle Tracking

• Capability to track individual vehicles
• Data and analysis ideas
• Speeds
• Acceleration / deceleration characteristics
• Lane changing behavior
• Turning behavior
• And more

36
Intersection Video
Unregistered Video orig_tuc.avi Registered
Video tuc_int_340.avi Registered
Video track_cars_cc.avi with Tracking
37
What does the future hold?

• Technology capabilities
• Methodology exists, is now being automated
• Some technology is mature, others are maturing
• Costs
• Automatic data reduction can cut costs
  significantly
• Equipment is dropping in price
• Imagery provides data on many traffic variables
• Airborne imagery may be competitive, especially
  in terms of cost per unit of traffic data