Login

Project

#25 Integrating transit signal priority with adaptive signal control in a connected vehicle environment


Principal Investigator
Stephen Smith
Status
Completed
Start Date
Jan. 1, 2016
End Date
Dec. 31, 2017
Project Type
Research Advanced
Grant Program
MAP-21 TSET National (2013 - 2018)
Grant Cycle
2016 TSET UTC
Visibility
Public

Abstract

This project aims at development of a more effective approach to transit signal priority (TSP) through integration of real-time adaptive signal control technology with DSRC-based detection of buses and communication of status information. We focus specifically on extending the Surtrac adaptive signal control system (in particular, the core intersection scheduling procedure) to achieve TSP objectives while at the same time minimizing the disruptive effects to overall traffic flow efficiency. Recent work with Surtrac has produced an intersection scheduling procedure that inputs weights reflective of the relative priority of vehicles and pedestrians, and uses these weights to generate signal timing plans that minimize the cumulative weighted delay of currently perceived incoming traffic flows. Taking this procedure as a starting point, we first investigate an extension that factors knowledge of bus stop locations and expected dwell times into Surtrac’s aggregate representation of traffic inflows to more accurately reflect bus arrival times at the intersection (as well as the arrival times of passenger vehicles that are likely to be blocked during dwell time at the bus stop). In Phase 2, we consider the added benefit of exploiting real-time bus status information, including such factors as bus schedule status (ahead or behind), bus occupancy, and bus door status (open or closed). In each of these two phases of the project, technology results will be evaluated first in simulation and then in the field. Field tests during both research phases will be carried out on the 24-intersection, DSRC-equipped test bed network that comprises the western end of the East Liberty Surtrac deployment.    
Description
Motivation: Transit vehicles are a major component of urban traffic flows, and create special challenges for control of signalized intersections. The traffic flow behavior expected from conventional fixed signal timing plans can be significantly disrupted by the presence of bus stops, resulting in loss of coordination across intersections, increased congestion and increased wait times. These effects, in turn, negatively impact the reliability of transit vehicle schedules, degrading the experience of the large number of travelers that depend on transit for mobility (which ultimately affects ridership levels). To attempt to keep on schedule, transit drivers often attempt to move through intersections during yellow and red phases, which creates an unsafe environment for pedestrians and other vehicles in addition to increased congestion. To address these issues, Transit Signal Priority (TSP) systems have been introduced in various urban environments. These systems operate by equipping transit vehicles with a device that communicates priority requests to a dedicated receiver at the intersection, which in turn responds to requests by overriding the installed timing plan to either cut a crossing green phase short or extend the current green phase and thus facilitate the bus’s movement through the intersection. These systems can substantially improve the movement of transit vehicles through congested urban road networks. However TSP systems also have several shortcomings: (1) unconditional priority is given to transit vehicle requests, to the potential detriment of all other vehicles moving through the intersection, (2) priority requests are issued by the driver, who is typically acting myopically to move the bus forward without consideration of impact to the existing traffic flows in the network, (3) in circumstances where competing priority requests are received, adverse effects to overall traffic flow are compounded by a basic first come, first served policy, and (4) there is no reason to push a bus forward if it is not behind schedule. Recent work on adaptive traffic signal control in urban environments, which has demonstrated the potential for significant gains in vehicle travel times, traffic throughput and air quality, provides a natural basis for satisfying TSP objectives in a more integrated manner that minimizes the adverse effects on overall traffic flows. By utilizing onboard communication with the intersection as a means of bus detection rather than a conduit for priority requests, and incorporating knowledge of bus stops and dwell time into the adaptive procedure for allocating green time to various intersection phases, we believe that active attention and priority can be given to buses within the broader framework of optimizing network-level traffic flows, and that TSP benefits such as greater transit schedule reliability, more effective clearance of buses through intersections, greater safety to pedestrians, etc., can all be achieved without undo degradation to overall traffic flow efficiency. We propose research aimed at providing this capability and demonstrating its effectiveness in the field. Technical Approach: Our recent work with the Surtrac adaptive traffic signal control system has produced an extended intersection scheduling procedure that aggregates sensed pedestrians and vehicles (and potentially other traffic modes) into a set of integrated, multi-modal traffic flows. Under this scheme, traffic flows are presented to the intersection scheduling procedure as sequences of aggregate multi-modal clusters (i.e., queues and platoons of vehicles and pedestrians). Each cluster in then weighted as a function of their constituent mode(s), allowing the generation of intersection timing plans that reflect mode priorities (as specified by the weights) while continually to adaptively balance overall traffic flows. The effectiveness of this procedure to better service pedestrians without undue sacrifice to vehicle traffic flow efficiency was demonstrated in simulation under the assumption that necessary pedestrian sensing capability was in place. We propose to build on this weighted intersection scheduling procedure to investigate the development of an adaptive signal control system that provides TSP capabilities while continuing to emphasize overall traffic flow efficiency. To enable detection of buses and determination of requisite location information, we will rely on vehicle-to-infrastructure communication via Dedicated Short Range Communication (DSRC) radios that are mounted both onboard the buses and at each intersection in the target road network. In the first phase of the project (2016), work will focus on developing a basic DSRC-based approach to adaptive signal control with TSP. The principal technical challenge here will be to extend the aggregate representation of traffic flows used by Surtrac to incorporate mode information (received via DSRC) and knowledge of the presence of bus stops and expected dwell times (to be additionally provided as part of an intersection configuration file). Specifically, bus location data provides a means of splitting clusters into sequences of sub-clusters that are either headed by a bus or contain only passenger vehicles. Knowledge of bus stops and dwell times can then be used to more accurately estimate cluster arrival times at the intersection. Equipped with this more refined representation of intersection inflows, the mode-waited intersection scheduling procedure can be used directly to achieve TSP. We will analyze and validate the developed approach, first using a VISSIM simulation model of the 24 DSRC-equipped intersections in Surtrac’s current Pittsburgh deployment (along Baum-Centre corridors) and then subsequently equipping Port Authority buses that run through the Baum-Centre corridors and performing a test of performance in the field. During the second phase of the project, work will focus on development of more advanced DSRC-driven adaptive TSP procedures that make use of real-time information about the status of an approaching bus to refine TSP control decisions. Port Authority buses are equipped with real-time sensing and information collection capabilities (provided by Clever Devices, Inc.). Potentially useful real-time information from a signal control perspective includes schedule status (so that priority can be conditioned on whether the bus is ahead or behind), how full the bus is (which could also suggest an adjustment to its cluster’s priority), and when the doors are opened and closed (which not only provides for the most accurate accounting of dwell time, but for near-side bus stops, it also gives a clear indicator that the bus is ready to leave the bus stop). As in Phase 1, an intersection scheduling procedure driven by such context-dependent prioritization will be tested and validated first in simulation and then in the field. The proposed project plan, which is outlined in the next section, assumes that additional complementary funding to this proposal will be obtained, and we are actively pursuing opportunities with our deployment partner: the Port Authority of Allegheny County. (In fact, there is a strong possibility that the Port Authority itself will provide this additional funding and this discussion is progressing) The present proposal is intended to fund the development, and validation of the algorithmic extensions to Surtrac necessary to achieve adaptive TSP signal control, and if we were successful this would constitute a significant scientific advance in its own right. The anticipated additional funding would support the acquisition of On-Board DSRC Units (OBUs), installation of OBUs on Port Authority buses, integration of DSRC-based sensing with Surtrac, and integration of Clever Devices’ onboard computer with the bus’s DSRC OBU (which would require some development effort by Clever Devices also) for validation of the approach in the field. If for some reason this additional funding is not obtained (which is deemed unlikely), then the level of effort allocated to developing more advanced versions of our adaptive TSP signal control system in Phase 2 will be reduced as necessary so that at least one field deployment can be undertaken.


[1] Smith, S.F. G.J. Barlow, X-F Xie, and Z.B. Rubinstein, “SURTRAC: Scalable Urban Traffic Control”, Proceedings 92nd Transportation Research Board (TRB) Annual Meeting, Washington DC, January 2013 [2] Xie, X-F., S.F. Smith, T. Chen and G.J. Barlow, “Real-Time Traffic Control for Sustainable Urban Living”, Proceedings 17th International IEEE Conference on Intelligent Transportation Systems, Qingdao, China, October 2014.
Timeline
Phase1: Basic DSRC-based adaptive signal control with TSP
? [1/1/16 – 4/30/16]   Task 1.1: Extension of aggregate traffic flow model to incorporate mode information and to account for expected delays at bus stops along the intersection’s entry road
segments.
? [5/1/16 – 8/31/16]   Task 1.2: Unit testing and comparative benefit analysis of current Surtrac procedure and extended Surtrac procedure with TSP (at different mode prioritization weight settings) using a VISSIM model of the Baum Centre corridors (and assuming the ability to detect buses and obtain bus location information).
? [5/1/16 – 8/31/16]   Task 1.3: Integration of Surtrac with DSRC-based bus detection capabilities (provided via DSRC OBU)
? [9/1/16 – 12/31/16] Task 1.4: Field test of basic Surtrac TSP solution on Baum-Centre corridors. Phase 2: Advanced DSCR-based adaptive signal control with TSP using real-time bus information
? [1/1/17 – 6/30/17] – Task 2.1 Extension of Surtrac intersection scheduling model to incorporate and base green time allocation on real-time bus information such as ahead or behind
schedule, doors open/closed, number of passengers on board, etc.
? [1/1/17 – 6/30/17] – Task 2.2 Development of DSRC interface to Port Authority’s onboard Clever Devices processor.
? [4/1/17 – 8/30/17] – Task 2.3 Unit testing and validation of extensions using VISSIM model (again under assumption that real-time bus information is available)
? [9/1/17 – 12/31/17] – Task 2.4: Field test and evaluation of Advanced Surtrac TSP solution
Strategic Description / RD&T

    
Deployment Plan
As indicated earlier, we plan to field test our integrated ‘adaptive signal control with TSP’ approach on our existing 24-intersection connected vehicle test bed that runs along the Baum Boulevard and Centre Avenue corridors within the current Surtrac deployment site. Together with our deployment partner we will equip a set of the Port Authority buses that travel through the Baum-Centre corridors. We will measure before and after travel flow efficiency (delay, travel times, idle time) for buses exclusively, using historical data collected by the Port Authority over the test period. We will also conduct a series of drive though runs on major routes through this network (both corridors and crossing), to enable comparison of overall traffic flow metrics for all vehicles with those obtained prior to introduction of the adaptive TSP scheme. We assume that if the field tests prove the approach to be effective, we will make the deployment permanent.
Expected Outcomes/Impacts
Overall, we expect to develop a more integrated approach to TSP that demonstrates significant performance improvement over the current Surtrac adaptive signal control procedure in both the travel flow efficiency of buses (in terms of metrics that include delay, travel times, and idle time) and on-time performance of buses (i.e., schedule reliability). We also expect comparable (if not better) performance with respect to the above metrics to standard TSP approaches, with significantly less disruption to overall travel flow efficiency of all vehicles through the target road network. Both our simulation analysis and our field-test evaluation plans (see next section below) will be aimed at confirming these claims.
Expected Outputs

    
TRID


    

Individuals Involved

Email Name Affiliation Role Position
isaack@cs.cmu.edu Isukapati, Isaac Robotics Institute Co-PI Faculty - Research/Systems
sfs@cs.cmu.edu Smith, Stephen Robotics Institute PI Faculty - Research/Systems

Budget

Amount of UTC Funds Awarded
$89760.00
Total Project Budget (from all funding sources)
$89760.00

Documents

Type Name Uploaded
Final Report 25_-_Mobility21-Phase1-Final_Report.pdf July 20, 2018, 9:03 a.m.

Match Sources

No match sources!

Partners

No partners!