PEDESTRIAN COLLISION WARNING DEMONSTRATION PROJECT Approach and Findings Presented by Harry Saporta Kelley Klaver Pecheux, PhD 2015 APTA Bus and Paratransit Conference May 5, 2015
Overview of Presentation Study team Study objectives Technologies assessed/tested Test approach Evaluation approach Findings
Study Team Study was funded through an FTA Cooperative Agreement Study was conducted by TriMet Applied Engineering Management Corporation (AEM) Portland State University (PSU)
Study Objectives Demonstrate the ability of various commercially-available turn warning systems to provide timely warning to pedestrians/cyclists that a bus is turning or pulling into/away from a bus stop. Determine the effectiveness of the turn warning systems at intersections and bus stops. Determine the benefit-cost ratios associated with the turn warning systems. Define the environmental parameters under which advance warning should be provided to pedestrians/cyclists at intersections and at bus stops. Assess the effectiveness of an innovative warning sign at one intersection.
Technologies Assessed/Tested Three commercially-available turn warning systems: Turn warning system 1 Spoken warning activated by rotation of steering wheel Included two LED strobe lights on either side of the bus Turn warning system 2 Spoken warning activated by rotation of steering wheel Turn warning system 3 Beeping warning activated through use of the turn signal Included directional LED headlights BUS blank-out sign Activated by a bus waiting to turn at an intersection (signal actuation) Concept for the sign was developed by AEM Test signs were designed and built by TriMet Signs were installed by the City of Portland A fourth commercially-available turn warning system was assessed but not tested Spoken warning activated through use of the turn signal
Study Approach Demonstration Test approach Evaluation approach
Test Approach 45 buses equipped with systems (15 buses with each of the 3 turn warning systems). All test buses call came out of one maintenance facility. Each day, test buses were placed on 5 pre-selected test routes: Pre-determined number of each system on each route (ranged from 2-4, generally 3, depending on the total number of buses on each route) One test route was replaced halfway through the test due to public complaints. BUS blank-out sign placed at both ends of one crosswalk at a busy intersection in downtown Portland. Test ran 7 months (March-September 2014).
Turn Warning System #1 - Pedestrians, Bus is Turning.
Turn Warning System #2 - Caution, Bus is Turning.
Turn Warning System #3 - Beeping Warning
Directional LED Headlights
BUS Blank-Out Sign
Evaluation Approach Bus operator surveys Daily surveys conducted 3 times per week for 13 weeks 436 completed surveys Comprehensive bus operator survey 208 completed surveys Pedestrian surveys Field intercept surveys administered at 5 intersections 454 completed surveys
Evaluation Approach (cont d) Bus operator focus groups 4 focus groups with 27 operators Pedestrian focus groups 3 focus groups with 27 pedestrians Analysis of pedestrian behaviors 80 hours of video data collected at 4 intersections Interviews with TriMet staff Cost-benefit analysis
Findings
Overview of Findings There were a number of common themes that emerged throughout the study. There was a (somewhat wide) range of feedback regarding perceptions, acceptance, and recommendations for improving the technologies representing mixed findings. Benefit-cost analysis
Common Themes Warning volume finding an appropriate volume level for the warnings Sensitivity of warning activation finding the right sensitivity setting for warning activation Warning - selecting the right warning type/content Application of the warnings determining when and where the warnings should activate
Warning Volume Getting the volume of the warnings right was an issue throughout the demonstration test. Early complaints from operators and residents led to volume adjustments. Noise complaints declined, but a growing number of operators reported in the daily surveys that the volumes had become too low to be effective. Overall lack of consensus in feedback: Operator responses showed that the too loud margin was significantly greater for systems with the spoken warning than for the system with the beeping sound. A majority of pedestrians surveyed did not find the warnings to be intrusive to the environment. Of those that did, more found the spoken warnings to be intrusive than the beeping warning. Participants in the pedestrian focus reported that the volumes of the spoken messages were acceptable once they were adjusted, but that the beeping warning was still too loud. Repetition and/or frequency of warnings may also play a role.
Sensitivity of Warning Activation Specific to the systems activated by rotating the steering wheel. Activation angles for both systems were initially set to provide early warnings. Initial turn angles produced false activations of the warnings in certain situations. Mitigated somewhat via system adjustments to one of the systems, which was evidenced by a downward trend in the reporting of false activations in the daily operator survey. A majority of operators still indicated in the comprehensive survey that the systems activated too early and/or in roadway curves. Operator focus groups confirmed that the problems were mitigated somewhat via system adjustments but continued in some cases. Speed threshold may also play a role.
Warning Type/Content Operators were most divided over the beeping warning: Some preferred it over the spoken warnings because it is universal and/or more effective at getting people s attention. Others did not like it because it was too loud, harsh, irritating, and potentially distracting. Strong consensus that caution is better than pedestrians. In addition to warnings used in the demonstration test, a variety of auditory warnings were presented to and discussed with focus group participants: Shorter, concise warnings were generally preferred over longer, more detailed warnings. Caution, bus is turning was rated above all others and was favored by most. A recommendation from both operators and pedestrians: Easily-recognized and unique to TriMet buses, as well as friendly. Warning that incorporates both a spoken warning and a sound/tone.
Application of the Warnings Overwhelming agreement by operators that the turn warnings are important, if not more important, at bus stops than at intersections (pedestrians tended to agree). Generally (slightly) more necessary during right turns (as opposed to left turns) and when pulling into a stop (as opposed to pulling away from a stop). Two recurring issues suggest that it may be prudent to consider selective versus ubiquitous application of the turn warnings: Early complaints - giving more thought to where the warnings should activate (and at what times of the day) could help to avoid these types of complaints. Long-term efficacy - bus operators and pedestrians believed that the warnings would eventually blend into the background noise and/or be tuned out. Both operators and pedestrians strongly recommended that the volumes vary by location and/or time of day or that the volumes adjust automatically relative to the environment. Alternatively (or additionally), participants recommended that warnings activate only on trouble routes/locations and/or at locations with history of conflicts. Flip side issue of liability if a pedestrian is struck at a location or time of the day when the warnings are inactive. Important to involve the community when selecting locations.
Wide Range/Mixed Feedback Regarding: Perceptions of technology effectiveness Acceptance of the technologies Recommendations for improving the technologies
Perceptions of Technology Effectiveness Overall, bus operators were generally less favorably impressed with the effectiveness of the turn warning systems than the general public: As is, the turn warning systems are only somewhat effective at improving safety (system and pedestrian related). Improvements, programming/customization, and some added operator control features could make the systems more effective. Systems not the be-all-end-all to reducing pedestrian-bus collisions (operator and pedestrian accountability). The warnings are as or more effective at service stops than at intersections, with pulling into a stop being the more critical movement. The turn warning systems are less effective with cyclists than with pedestrians. LED cornering headlights highly effective at improving visibility in the direction of the turn. A fair majority of pedestrian respondents felt that the systems were effective at both alerting pedestrians and improving pedestrian safety. Just over half of pedestrian respondents found the BUS blank-out sign to be effective at alerting pedestrians that a bus is turning and at improving pedestrian safety.
Acceptance of Technologies Operators: Nearly half of operators surveyed agreed that the potential safety benefits outweighed the drawbacks of the warning systems. However, overall, only about one third agreed with the prospect of wider deployment, while nearly half disagreed. General sentiment from focus groups regarding further deployment could be described as ranging from apathetic to skeptical most seemed to have adapted to the presence of the warnings, but did not strongly support nor completely reject the idea of their continued or expanded use. Pedestrians: A majority of pedestrians surveyed did not find the warnings to be intrusive to the environment. Further, a majority agreed that more systems should be installed. A small group agreed that the potential benefits outweighed the associated drawbacks, but disagreed with the idea of further deployment. Could be associated with costs money better spent elsewhere.
Recommendations for Improving Technologies Tweak the systems and get the bugs out. Tie the system to the turn signal (as opposed to the steering wheel). Select a lower speed threshold for warning activation. Integrate systems with GPS/AVL system to be more programmable. Operator control Most operators agreed that total operator control over the system was not a good idea. Some operators advocated for some level of control (e.g., volume and when warning could be sounded). Others strongly recommended that operators be given no control over the system. If more programmable, would reduce/eliminate need for any operator control.
Costs and Benefits Imputed benefits Cost-benefit analysis
Imputed Benefits Framework Bus-pedestrian collisions resulting in injuries or fatalities are rare events. 2010 National Transit Database 27 fatalities and 283 injuries requiring transport in association with 1.6 billion miles of revenue service. This translates to about 1.7 fatalities and 17.7 injuries per 100 million revenue service miles. TriMet s entire bus system logs approximately 20 million revenue service miles per year. A research design predicated on documenting changes in fatalities and injuries on six routes over a 7-month period would likely find no treatment effect attributable to the turn warning systems. Drew on safety pyramid concept Bus-pedestrian close-calls more common than actual collisions. Given a defined hierarchy of safety risks and outcomes, it is possible to impute the expected incidence of pedestrian fatalities and injuries in relation to the frequency of close-calls. Employed relative frequency data specific to close-calls and collisions involving buses and pedestrians transit property data) to operationalize pyramid for this study. (national and the safety
Imputed Benefits National data 2010 NTD Fatalities and injuries requiring transport Transit property data TriMet Safety information system (Dec 10-Oct 14) Bus-pedestrian collisions, injuries (requiring transport minor, none) Close-calls Pedestrian close-call button (since 2013) not well documented (under reported) by operators Data from 2012 survey of risk perceptions frequency of total close calls Evasive action/hard stop events due to pedestrians Daily operator surveys reported reduction in close calls due to turn warning systems 1.0 10.5 30.0 37,723 Incident Types Close-Calls per Incident Annual Avoided Incidents Years to Realize Fatality 37,723 0.037 27 Injury (transport) 3,593 0.393 2.5 Minor or No Injury 1,257 1.124 11 months
Cost-Benefit Framework Guidance from OMB Circular A-94 Information on discounting Treatment of uncertainty Treatment of the Economic Value of a Statistical Life in Departmental Analysis, U.S. DOT Monetary valuation information 12-year cost-benefit analysis timeframe 7% discount rate
Cost Analysis Accounts for uncertainty by employing a range of values for each key parameter: Avoided incidents Valuation of fatalities and injuries Installation and maintenance costs Three scenarios Baseline - employs the baseline cost and benefit component values Maximum scenario employs the combination of cost and benefit values that yield the largest achievable net benefit outcome Minimum scenario employs component values that produce the smallest net benefit
Results Cost-Benefit Components Baseline Maximum Minimum A. Present Value Total Benefits $3,047,131 $7,055,243 $693,591 B. Present Value Total Costs $108,669 $66,205 $151,134 C. Net Present Value (A-B) $2,938,462 $6,989,038 $561,807 D. Ratio of Benefits to Costs (A/B) 28.0:1 106.6:1 4.6:1 E. Internal Rate of Return 34.5% 51.4% 16.5% Results are for the 45 test buses on the 6 test routes. Potential elevated relative risk exposure: While representative of TriMet s overall network, the test routes are somewhat longer, have more turns, and serve more passengers than average. Systems were installed on new buses assigned to base service runs. Monetary benefits from avoided fatalities/injuries due to the turn warning systems include social monetary benefits as well as avoided financial liabilities realized by a transit agency.
Questions? TriMet Harry Saporta Safety and Security Executive SaportaH@trimet.org (503) 962-4909 AEM Corporation Kelley Klaver Pecheux Associate Director of Transportation Kelley.Pecheux@aemcorp.com (703) 464-7030 ext 8117 Federal Transit Administration Roy Wei-Shun Chen Transportation Engineer RoyWeiShun.Chen@dot.gov (202) 366-0462