Improve passenger information accuracy and reduce redundant hardware without overwhemling communications capacity, by integrating adaptive traffic signal priorty and dynamic passenger information systems into existing AVL/ACS systems on transit vehicles.

Results of a California PATH test project in San Mateo County, California.

Date Posted
01/31/2013
Identifier
2013-B00818
TwitterLinkedInFacebook

Development of an Integrated Adaptive Transit Signal Priority (ATSP) and Dynamic Passenger Information (DPI) System

Summary Information

The transit industry has applied various advanced Intelligent Transportation System (ITS) technologies to improve the efficiency of transit operations and to provide passengers more reliable transit service in order to better serve their riders. Among these ITS technologies, the Automatic Vehicle Location and Advanced Communication System (AVL/ACS), Transit Signal Priority (TSP) and Dynamic Passenger Information (DPI) are widely deployed technologies. This research is a continuation of two previous California PATH research projects on the development and implementation of an adaptive transit signal priority (ATSP) system. This research project was conducted to develop an integrated ATSP/DPI system that utilizes the existing hardware and communication link offered by AVL/ACS. The eventual goal is to have one integrated system to support the purposes of fleet management, TSP and DPI. The hypothesis was that the integration approach is not only cost effective, but can also improve the performance of dynamic passenger information. When integrated with ATSP, DPI processes share the information regarding ATSP operation status leading to better predicted bus arrival times at bus stops.

Methodology
The ATSP/DPI was integrated with a cell phone based emulation of the ACS since this was only a test project and integration with the actual ACS system was not feasible. This system was field tested on SamTrans routes 390 and 391, of which the central portion is ATSP enabled. Outcomes were analyzed based on a four week "before" period and an eight week "after" period.

Findings
During the eight weeks of field testing, the additional communications required for incorporating ATSP/DPI into the AVL/ACS system were handled without overwhelming the channel capacity of the ACS. On average about 120 polls were dynamically inserted along the 10-mile-long ATSP corridor for the average travel period of 55 minutes. The average time gap between inserted polls was 27 seconds. Ninety-seven percent (97%) of inserted polls were processed without the collision of multiple competing polling requests. Even when a request collision did happen, the delay in responding to inserted polling request was within 4 seconds.

Dynamic passenger information based on the integrated ATSP/DPI was displayed to the public at Millbrae Station kiosks.
  • Prediction error of bus arrive time was within +/- 2 minutes when the bus was 40 minutes away.
  • SamTrans personnel observed site multiple times and found the information "is remarkably accurate".
  • Integrating DPI with ATSP improved the performance of DPI, as the inserted polls for the purpose of ATSP contributed to better prediction results.
  • The absolute error mean was reduced by 16% in the section covering four time-points within the ATSP corridor.
Goal Areas
Deployment Locations