Integrated Corridor Management Systems Impacts of Data Quality on - PowerPoint PPT Presentation

Integrated Corridor Management Systems Impacts of Data Quality on Decision-Making Dr. Francois Dion Dr. Qijian Gan University of California - Berkeley 04/25/2019

Outline 2  Definition of ICM Systems  Example: I-210 Pilot ICM System  Data Needs and Issues

Definition of ICM Systems 3

Integrated Corridor Management Systems 4  Systems attempting to bring together the operation of:  Freeways  Arterials  Transit systems  Emergency vehicles  Parking systems  Greatly focus on the use of technology and information processing to:  Monitor the operation of various systems  Provide value-added information to system operators  Provide relevant information to travelers

Project Goals 5  Typical project goals  Improve operational situational awareness  Promote collaboration among corridor stakeholders  Improve incident response  Improve travel reliability  Improve overall corridor mobility  Empower travelers to make informed travel decisions  Facilitate multi-modal movements across the region  Improve corridor safety

Example: I-210 Pilot ICM System 6

The I-210 Pilot ICM 7  Freeway/arterials projected to be under ICM management EB Detours WB Detours

System Integration 8 Freeway Detectors Lane Closure System CAD/AVL System Ramp & Connector Meters Foothill Transit Caltrans TMC ATMS Freeway CMSs Traffic Signals - TSMSS CAD/AVL System Traffic Signals - SCATS Pasadena Transit Traffic Signals - Transparity Travel Time Devices Rail CAD/AVL System Pasadena TMC Arterial CMSs Bus CAD/AVL System Electronic Wayfinding Signs LA Metro Park & Ride Management System Traffic Signals - TransSuite Arcadia TMC Travel Time Devices ICM Parking Parking Management Electronic Wayfinding Signs Operators Systems ( future ) DSS DSS Traffic Signals - KITS California County TMC Travel Time Devices CAD System Highway Patrol Electronic Wayfinding Signs Traffic Signals - KITS Local Law Monrovia Local Dispatch Systems Travel Time Devices Enforcement Electronic Wayfinding Signs Freeway Service Patrol Duarte LA SAFE Traffic Signals - KITS 511 System Travel Time Devices Call Boxes Electronic Wayfinding Signs Information Other Data Providers Suppliers PeMS Traveler Information Apps Probe Vehicle Data Radio Crowd Sourcing Weather

Supporting Simulation Model 9

Data Needs and Issues 10

Data Needs: Estimation of Traffic Conditions 11  Current traffic conditions Items difficult to obtain • Adequate estimation of  Freeway segments existing traffic demand ◼ Mainline lanes • Arterial traffic detection at a finer granularity ◼ On-ramps / Off-ramps • Real-time signal control ◼ HOV lanes information  Arterial segments • Probe data with adequate ◼ Midblock counts penetration rates. ◼ Turning counts at intersections Impacts of inaccurate counts: ◼ Detector counts at intersections • Inadequate estimation of existing traffic demand • Misidentification of congested segments/areas

Data Needs: Management of Incidents/Events 12  How can we quickly obtain reliable incident Potential Impacts information? • Inadequate representation of  Incident/event characteristics (location, lane(s) existing conditions affected, duration) • Inadequate estimates of traffic demand  How can we identify unusual traffic patterns? • Development of solutions to address  How to quantify deviation from normal wrong problem conditions • Incorrect prediction of impacts of proposed  How can we track driver responses to response plan unusual traffic patterns? • System-level impact  Percent traffic exiting freeway from navigation app providers  Alternate route followed

Data Issues – Freeway Counts 13  Reliance on loop or point detectors only provide Potential Impacts information at specific locations • Inadequate  Must extrapolate what happen between detection stations representation of existing conditions  Poor detector health at certain key locations  Require additional work to either fix problematic detectors or use of other data sources as a supplement  Consistency of counts across successive detectors  Traffic counts at individual detectors do not always match upstream counts  Count accuracy in congested areas  Detectors may have difficulty counting vehicles in dense traffic  Counts reflect capacity, not necessarily the traffic demand

Data Issues – Freeway Counts 14  Inadequate coverage of entry/exit movements at Potential Impacts HOV gates • Inadequate  Need to be able to track vehicle movements with representation of lane-specific accuracy existing conditions  Wrong setup configuration between HOV and general lane detectors in cabinet  Require advance techniques to identify such problems

Data Issues – Arterials Counts 15  Counts on arterials are inherently very variable Potential Impacts  Effect of actuated/adaptive signals • Inadequate  Vehicle entering/exiting between main intersections representation of existing conditions  Ease of changing path to go around incidents or congested areas • Inadequate estimates of traffic  Non-ideal placement for accurate counts demand  Location of approach detectors may allow vehicles to go • Development of by without being detected solutions to address wrong problem  Incomplete detector coverage for traffic estimation  Some intersection approaches are only covered by either • Incorrect prediction stop line or advance detectors of impacts of proposed response  Count accuracy in congested areas plan  Counts from stop line detectors are normally underestimated when traffic is congested

Data Issues – Arterials Counts 16  How to measure turning counts from shared lanes? Potential Impacts  Possible, but typically requires more expensive detection • Inadequate system representation of existing conditions  How can we track vehicle movements? • Inadequate  Require data at a finer granularity, either event-based estimates of traffic data or probe data demand • Development of  Capacity reduction caused by shared lane solutions to address configuration and pedestrians? wrong problem  Throughputs at certain lanes are much lower than the • Incorrect prediction normal saturation flow rates in such a case of impacts of proposed response plan

Data Issues – Demand Modeling 17 How to model traffic demand?  Potential Impacts  Representative day? • Difficulty of  Representative weekday and weekend day? estimating true  Set of representative days: Mon, Tue, Wed, etc.? benefits of proposed actions  Set of representative days for each month?  How about holidays? How to capture impacts of incidents occurring outside  control area?  Outside incidents can affect traffic demand How to determine actual demand in presence of congestion?   Sensors measure roadway capacity, not actual traffic demand How can we determine routes followed by vehicles?   Probe data is promising, but current market penetration rates very low

Data Issues – Decision Support Systems 18 Due to processing needs, data used in decision support  Potential Impacts systems are generally a few minutes old • Control needs to be  Result in systems generally responding what has been based on trend observed several minutes ago • Need to anticipate Effect of navigational apps  lag in response  Navigational apps may push vehicles to take different route than plan by traffic managers Length of decision interval   Do we update control decisions every 5 min, 10 min, 15 min? Forecasting future traffic is possible, but may be affected  by various elements  Quality of the underlying data  Size/complexity of network  Normal fluctuations in traffic demand/patterns  Incidents

Data Issues - General 19  Consistency of data formats Potential Impacts  Various systems use different data formats, not • Inability to fuse always fully compatible together data from  Which format/standard to use? various sources

Conclusions 20  Current technology does not provide perfect data  Control decisions must account for  Potential data inaccuracies  Data collection lag  Drivers not behaving as desired  Apps providing contradictory directions