Estimating Vehicle Trajectories on a Motorway by Data Fusion of Probe and Detector Data

Transcription

1 Estimating Vehicle Trajectories on a Motorway by Data Fusion of Probe and Detector Data International Workshop on Transport Networks under Hazardous Conditions March 1st, 2013 Masao Kuwahara, Takeshi Ohata, Tsubasa Takigawa, Tohoku Univ. Koichi Abe, East NEXCO, Ltd. Takeshi Imai, Honda Motor Company Ltd. Kiichiro Nakamura, Metropolitan Exp. Co. Loop Detector (Intercity Motorways) e.g. One loop detector between interchanges (approximately one in every 10 km) Ultra sonic Detector (Urban Expressway, Local Streets) e.g. Metropolitan expressway Ultra sonic detectors installed at every 300 m 1

2 Probe Data GPS + Communication equipment GPS DMCA Mobile Infrared Microwave (DSRC) (Dynamic Multi Channel Access) Taxi Auto company Police agency MLIT (Ministry of Land Infrastructure and Transport) Transport Centre Data Fusion Probe data:sampled data but spatially cover a wide area Detector data:measure all cars but at limited locations space detector probe probe detector traffic in the dashed area could be monitored by data fusion detector time 2

3 Richness of Probe Data Space Probe trajectory: has more to offer! Traffic conditions & driving modes What is used in practice: travel time only! Complete stop Free-flow Stop and go Acceleration Deceleration Time 6 Objective Space Downstream Departing vehicles Red intervals Probe trajectory Upstream Arriving vehicles Time 7 3

4 Kinematic Wave Theory and 3D Representation Lighthill Whitham Richards Newell Daganzo Three Dimensional Representation Cumulative Curves and Time Space Diagram (vehicle trajectories) space time 4

5 Estimating 3D Cumulative Surface Cum.Trips Forward Wave Exit Study Section Probe Entry Backward Wave Time 10 Estimating 3D Cumulative Surface Cum.Trips Forward Wave Exit Study Section Probe Entry Backward Wave Time 11 5

6 Probe + Detector Pulse + Signal Timing Detector Estimate vehicle trajectories of all running vehicles FLow Signal FD Density Travel time variability Environmental evaluation Signal control improvement Probe Detector 12 Test bed Komazawa Street 13 6

7 Probe + Detector Pulse + Signal Data Fusion Space [m] Detector Signal Timing Probe trajectory Detector Probe data have very rich information, not just travel time. Fusing probe information with other data, a complete picture of traffic condition can be drawn. time Fundamental Diagram 7

8 Complete Trajectories based on Kinematic Wave Theory without Probe data Big gap Estimated trajectory Probe trajectory without Probe data (only vehicle passing times at detector locations and signal timings are used) Complete Trajectories based on Kinematic Wave Theory 4 th Probe (1 second freq) with 4th Probe data (1 second uplink frequency) (together with vehicle passing times at detector locations and signal timings) 8

9 Consideration of Incoming & Outgoing Vehicles in the middle of the Study Section 18 Incoming/Outgoing Vehicles Influential Area in Time-Space Suppose a vehicle enters (or leaves) at A Cumulative height at A influences only in Region I II III A I Space Time u IV 19 w i 19 9

10 Incoming/Outgoing Vehicles Simultaneous Calculation: Adjusting Link Costs B Unchanged C II Space III Time u A IV I +1 w i D E +1 in advance 20 A modified trajectory by adding and removing incoming and outgoing vehicles 21 10

11 Measured and Estimated Probe Trajectories Application to Intercity Motorway Tohoku Exp Inbound:Shiroishi IC to Motomiya IC August 14,

12 Probe Trajectories and IC Locations 24 Estimated Trajectories using One Probe without Considering In/Out Vehicles 25 12

13 Estimated Trajectories using Five Probes without Considering In/Out Vehicles 26 Estimated Trajectories using Five Probes Considering In/Out Vehicles 27 13

14 RMS in relation to Entrance Time Summary Estimating trajectories of all running vehicles on an arerial with intersections and on an intercity motorway Fusing probe data with detector and signal timing data Emphasizing the utilization of rich information of probe trajectories Variational Theory oriented from Kinematic Wave 29 14