Novel Vehicle Mass-Based Automated Passenger Counter for Transit Applications. Issue 1 (January 2015)

• Record Type: Journal Article
• Authors: Kotz, Andrew J.; Kittelson, David B.; Northrop, William F.
• Is Part Of: Transportation research record. Volume 2563:Issue 1(2016)
• Page Start: 37
• Page End: 43
• Publication Date: 2015-01
• Subjects: Transportation -- Periodicals; Roads; Transport -- Périodiques; Routes -- Périodiques; Routes -- Conception et construction -- Périodiques; Roads; Transportation; 388.05
• DOI: 10.3141/2536-05
• Languages: English
• ISSNs: 0361-1981

Abstract

Federally subsidized transit funding is based on ridership; therefore, an accurate count of passengers is imperative. Most existing automated passenger counters (APCs) use infrared beam methods of detection. Such systems are expensive and inaccurate for scenarios such as multiple-passenger boarding and alighting. The preliminary results are reported here from a novel APC method that has the potential to improve accuracy over existing technology while decreasing overall system cost. This result is accomplished through integration of existing vehicle systems that include the vehicle air ride suspension, which has near-universal adoption in transit buses. The system counts passengers by measuring pressure inside the vehicle air bag suspension system, which directly correlates to vehicle mass. Two algorithms were developed to detect discrete boarding events on the basis of time-resolved vehicle mass data. The first algorithm uses incremental change of vehicle mass, assuming a well-calibrated average passenger mass of 168 lb (76 kg), resulting in a −2.4% error. The second algorithm uses the vehicle mass time derivative to improve the resolution of boarding and alighting events but with overall error increased to −28.4%. Experimental testing of the mass-based APC system on an in-service transit bus showed that the mass correlation method outperformed the existing infrared beam APC, which had a 17.5% error; however, bus kneeling events proved problematic for both the mass correlation and the event-based mass methods. Initial results are encouraging and prompt the necessity for further study and refinement. However, more work must be done to address the kneeling issue.