Development of a Low-Cost Automatic Vehicle Classification (AVC) System (original) (raw)

Abstract

The main objectives of this research were to review and evaluate the current automatic vehicle classification systems and operating procedures and to design and recommend a modified AVC system that utilizes weights for improved performance. Recommendations: tubes should be used for all AVC installations as they typically had a lower frequency of missed axles/vehicles compared to piezoelectric sensors. Tubes should be installed about 6" from the edge of the inside lane marking, which would minimize the potential for stray vehicles from an adjacent lane to activate the sensors. Use of inductive loops generally would not be advantageous, except for relatively congested facilities and should be used in urban facilities that experience recurring congestion. The field test results show that the overall accuracy for all vehicles was greater than 90%, although accuracy for trucks was generally site specific. Using a different binning scheme for different sites might improve performance on individual sites. Typically, about 1 to 4% of the vehicles were missed, thus not classified by the system. Passenger vehicles are generally most affected, and proportion was typically higher when a loop was used.

Shashi Nambisan hasn't uploaded this paper.

Let Shashi know you want this paper to be uploaded.

Ask for this paper to be uploaded.