Highway Transportation Management Systems

1
Highway Transportation Management Systems
Vehicle Classification Methods and ORT Vehicle
Classification
By Brian Patno
2
Vehicle Classification Methods

• Weight
• Weight per axle
• Number of Axles
• 2 axle configurations can vary significantly
  which usually requires a height component for
  some axle configurations
• Sometimes single or double tire widths are also
  measured
• Vehicle Class
• Determined by Vehicles length, width, and height.

3
Weight Systems

• Positives
• Provides a critical measurement to determine the
  safety of a commercial vehicle
• Negatives
• Requires multiple sensors
• Requires a sensor in the road
• Maintenance actions can cause road or lane
  closure
• May require a vehicle to stop or slow down

4
Axle Counting

• Positives
• Traditional vehicle classification method used
  around the world
• FHWA approved Vehicle Classification (Scheme F
  has 13 vehicle classifications)
• Negatives
• Requires multiple sensors
• Requires a sensor in the road
• Maintenance actions can cause road or lane
  closure
• May require a vehicle to stop or slow down

5
Vehicle Classification Technology
6
Vehicle Classification Technology
Next Generation Approach
Classification while promoting maximum capacity
and without impacting road users
Revenue neutral to traditional classification
methods
Vehicle Class
Lasers
Video
7
ORT Classification Requirements

• No restriction on Vehicle movement
• Not required to stop or slow down
• Not required to be in a lane
• Does not require in the road mounting
• All weather performance
• High Vehicle Classification accuracy
• Detection Requirements
• Provides accurate vehicle location information
• High Probability of Vehicle Detection with low
  False Alarm Probability
• Provides highly accurate and precise Camera
  Trigger

8
Stereoscopic Cameras

• Requires good illumination of the vehicle
• Large Field of View to capture entire vehicle
• Computation intensive

9
Infrared lasers

• Extensive Computation
• Requires good reflection of energy back to the
  detector.
• Provide range and intensity

Intensity Image
Range Image
10
Extensive Computation means

• Use probability density functions (PDFs)
• Decompose an image using wavelets or other
  methods
• Distinguish targets from their background clutter
  
• Algorithms are adapted or trained using the tools
  of experimental methodology to determine the
  algorithm parameter sets
• Then the detection probabilities as functions of
  false alarm are estimated or calculated using
  previously acquired data sets
• Some example parameters target pixel size,
  target-to-background noise ratio, target to
  background interference ratio

11
Detection and False Alarm are interdependent
All Sensors follow this fundamental principle
S/N
Low S/N increases PFA
PD 99.9 with PFA10-12 requires a S/N of 20 dB.
PD
12
Two separate beams improve performance
Raytheon patented the message sequence to
determine vehicle detection, trigger camera and
provide vehicle classification
13
Some Conclusions

• Open Road Tolling should be used to maximize road
  capacity and minimize road closure
• No restrictions or impacts on toll road users
  caused by the tolling technology
• Overhead sensors satisfy this requirement
• Raytheon has not had any safety accidents with
  over 7 years of operation and maintenance using
  sensors mounted above the toll road
• Overhead sensors can separate vehicles into the
  toll classes with an accuracy exceeding 90, but
  need more work to achieve the desired 99
  accuracy with very low over or under
  classification errors