Real-time Microscopic Estimation of Freeway Vehicle Positions from the Behaviors of Probe Vehicles Noah J. Goodall, P.E. Research Scientist, Virginia Center for Transportation Innovation and Research Ph.D. Candidate, University of Virginia April 20, 2012 10/27/2019
Smartphones Very sophisticated computer Sensors GPS 3-axis accelerometer Camera Magnetometer Carried with you all day
Modern Vehicles – Very Sophisticated How many lines of code in a: F-22 Raptor: Average new Ford: 1.7 million 10 million
Computerized Measurement Speed Heading Acceleration (lateral, longitudinal, vertical) Position (from GPS) Other diagnostics Wipers on/off Headlights on/off Braking status Turn signals on/off Tire pressure Rain sensors Steering wheel angle Stability control
Vehicle-to-Vehicle Communication: Not Sophisticated Hi-tech vehicles Low-tech communication with other vehicles Brake lights Turn signals Horn Flash headlights
Vehicle-to-Infrastructure Communication: Not Much Better We want to know where vehicles are, what they’re doing Lot’s of sensors already in the field to detect this
Field Sensor Shortcomings Limited data quality Point detection, not continuous coverage Difficult/expensive to repair = frequent downtime Limited types of data Aggregated speed, density, and volumes at a single point
Solution: Connected Vehicles
Wireless Vehicle Communication Significant movement towards wireless communication between vehicles and infrastructure
Connected Vehicle Applications Lots of connected vehicle mobility applications in development Most of these applications need at least 20-30% of vehicles to be “connected” to see benefits Application % Connected Vehicles Needed for Benefits Traffic signal control 20-30% Freeway incident detection 20% Lane-level speed estimation Arterial performance measurement 10-50%
Better Performance with Higher Market Penetration Premier and Friedrich, “A Decentralized Adaptive Traffic Signal Control Using V2I Communication Data,” Proceedings of the 12th International IEEE Conference on Intelligent Transportation Systems, October 2009.
Background Rollout of connected vehicles will not be instantaneous 16 years between kickoff and 80% penetration Projected rollout of on-board equipment in US Fleet (Volpe, 2008)
Assumed Location of Unequipped Vehicle What it Means Problem – Not able to get the full benefit of connected vehicle applications at low market penetrations. Solution – “Location Estimation” Behavior of equipped vehicles may suggest location of unequipped vehicles. Assumed Location of Unequipped Vehicle Equipped Vehicles
Methodology How to estimate vehicle locations Depends on unexpected behavior of equipped vehicles – indicates an unequipped vehicle ahead What is “unexpected”? Car-following model
Algorithm Vehicles assumed to follow Wiedemann car-following model Widely accepted, basis for VISSIM A deviation from expected acceleration indicates an unequipped vehicle ahead -4 ft/s2 actual Estimate properties from model or history Wiedemann is widely accepted 3 ft/s2 expected 30 mph 45 mph Headway = 97 feet Speed = 29 mph
Algorithm Vehicles assumed to follow Wiedemann car-following model Widely accepted, basis for VISSIM A deviation from expected acceleration indicates an unequipped vehicle ahead -4 ft/s2 actual Estimate properties from model or history Wiedemann is widely accepted 3 ft/s2 expected 30 mph 45 mph Vehicle continues to drive according to model, until overtaken 10/27/2019
Testing Using NGSIM datasets as ground truth Two short freeway segments Artificial VISSIM model to supplement One direction, 4 lanes, 7 miles, 2 interchanges
Results Equipped Observed Predicted
I-80 at 25% Market Penetration Ground Truth Densities (Equipped and Observed Vehicles) Distance (1/8 mile) Estimated Densities (Equipped and Predicted Vehicles) Distance (1/8 mile) Time (s)
Artificial Network at 25% Market Penetration Ground Truth Densities (Equipped and Observed Vehicles) Distance (7 miles) Estimated Densities (Equipped and Predicted Vehicles) Distance (7 miles) Time (s)
Performance Effective Market Penetration = Accurate Predictions – False Predictions + Equipped Vehicles Total Actual Vehicle-Seconds The blue line comes in early.
Conclusions The algorithm can predict the locations of many unequipped vehicles at various levels of accuracy, particularly during and after congestion
Next Steps Apply on surface streets Vehicles decelerate for many reasons signals, pedestrians, stop signs, turning, etc. Lots of congestion Almost every vehicle queues at some point
Special Thanks Advisors Brian L. Smith, P.E., Ph.D., University of Virginia Byungkyu (Brian) Park, Ph.D., University of Virginia
For more information: Noah Goodall noah.goodall@vdot.virginia.gov