Automated pedestrian detection at signalized intersections tested in three U.S. cities reduced the number of pedestrians who began crossing during the steady DON’T WALK signal by 81 percent.
Date Posted
01/01/2004
Identifier
2004-B00276
TwitterLinkedInFacebook

Evaluation of Automated Pedestrian Detection at Signalized Intersections

Summary Information

The objective of this evaluation was to determine if automated pedestrian detectors, when used in conjunction with manual push-button pedestrian crosswalk signal systems could decrease the likelihood of inappropriate crossings (i.e., beginning to cross during the DON’T WALK signal) and reduce vehicle-pedestrian conflicts.
 
Four intersections were monitored in three cities (Los Angeles, Rochester NY, and Phoenix) using video cameras to record pedestrian and motorist behavior before and after automated pedestrian detection systems were installed at busy crosswalks.  Preexisting manual push buttons were available at each intersection before and after the automated systems were activated.
 
In Los Angeles, both microwave and infrared detection systems were installed.  Infrared detectors monitored pedestrians moving in the crosswalk, and microwave detectors monitored pedestrians waiting at curbside queuing areas.  This system automatically called for a pedestrian crossing signal if a person was detected in the queuing area for a specified amount of time. In addition, the system delayed the green light for opposing traffic if a person was detected in the crosswalk during the DON’T WALK signal.  If a pedestrian was moving in the crosswalk at the end of a crosswalk clearance interval, the system automatically extended the crossing time by 0.2-second increments to a maximum of 6 additional seconds.

The Rochester and Phoenix sites were of different design. They deployed microwave type systems and restricted coverage to curbside queuing areas.
 
Data were collected using video cameras before and after the deployment to monitor pedestrians, vehicles, crossing signals, traffic signals, and push buttons. The following intersections and crosswalks were included in the evaluation project.
 

  • Los Angeles - San Pedro and Temple (East side crosswalk).
  • Rochester - Crittenden and Lattimore (North side crosswalk).
  • Rochester - State and Corinthian (North side crosswalk).
  • Phoenix - Central and Earll (South side crosswalk).

 
In Los Angeles, a camera was setup at the top of a parking deck, and in Rochester and Phoenix, cameras were setup on sidewalks located about 75 feet from each intersection.  The author noted that the video data were collected during dry weather conditions (i.e., no ice or rain), and the conditions and timing of each data collection period could have affected the results. 

The number of pedestrians and hours of data collected at each test site were detailed as follows.
 

STUDY SITE
BEFORE
Number of Pedestrians and Hours of Data Collection
AFTER
Number of Pedestrians and Hours of Data Collection
LA (San Pedro and Temple) 573 (2 hr) 533 (4 hr – w/ infrared)
590 (1 hr 50 min – w/ microwave)
Rochester (Crittenden and Lattimore) 767 (4 hr) 555 (2 hr)
Rochester (State and Corinthian) 277 (7 hr 20 min) 208 (3 hr 30 min)
Phoenix (Central and Earll) 500 (2 hr) 671 (4 hr)


(Source: Table 1 in Hughes, et al.)

RESULTS

Overall, after pedestrian detection systems were activated at each crosswalk, there was a 24 percent increase in the number of pedestrians who began crossing during the WALK signal, and an 81 percent decrease in the number of pedestrians who began crossing during the steady DON’T WALK signal.
 
The analysis indicated the addition of automated pedestrian detection to sites with existing pedestrian push-buttons will decreased the likelihood pedestrians will encounter opposing traffic at crosswalks. It should be noted, however, that the number of sites upon which these results are based is small and as the data in the report indicated, pedestrian performance can vary widely across sites. The report indicated that in the future, additional data should be collected at a larger number of sites to help traffic engineers analyze site selection and determine the types of locations where automated detectors are most effective.

Goal Areas