TRAFFIC PREDICTION USING REAL-WORLD TRANSPORTATION DATA

• US 20140114556A1
• Filed: 10/22/2013
• Published: 04/24/2014
• Est. Priority Date: 10/23/2012
• Status: Active Grant

• Current Assignee: University of Southern California
• Original Assignee: University of Southern California
• Inventors: Demiryurek, Ugur, Shahabi, Cyrus, Pan, Bei
• Granted Patent: US 9,286,793 B2
• Time in Patent Office: Days
• US Class Current: 701/119
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G08G 1/00 Traffic control systems for road vehicles arrangement of road signs or traffic signals E01F9/00 ; automatic vehicle control B62D
• CPC Class Codes: G08G 1/0112 from the vehicle, e.g. floating car data [FCD]
• CPC Class Codes: G08G 1/0129 for creating historical data or processing based on historical data
• CPC Class Codes: G08G 1/0141 for traffic information dissemination

Abstract

Real-time high-fidelity spatiotemporal data on transportation networks can be used to learn about traffic behavior at different times and locations, potentially resulting in major savings in time and fuel. Real-world data collected from transportation networks can be used to incorporate the data'"'"'s intrinsic behavior into a time-series mining technique to enhance its accuracy for traffic prediction. For example, the spatiotemporal behaviors of rush hours and events can be used to perform a more accurate prediction of both short-term and long-term average speed on road-segments, even in the presence of infrequent events (e.g., accidents). Taking historical rush-hour behavior into account can improve the accuracy of traditional predictors by up to 67% and 78% in short-term and long-term predictions, respectively. Moreover, the impact of an accident can be incorporated to improve the prediction accuracy by up to 91%.

Claims

1. A method comprising:

receiving a request relating to traffic prediction, the request having an associated day and an associated time;

determining how much to apply each of a first traffic prediction model and a second traffic prediction model based on previously recorded traffic data corresponding to the associated day and the associated time, wherein the first traffic prediction model comprises a moving average model that exhibits increased prediction accuracy as a prediction time horizon is reduced, and the second traffic prediction model comprises a historical average model that exhibits similar prediction accuracy across multiple prediction time horizons; and

applying the first and second traffic prediction models in accordance with the determining to generate an output for use in relation to traffic prediction.

2. The method of claim 1, wherein the determining comprises:

calculating a first prediction error for the first traffic prediction model and a second prediction error for the second traffic prediction model; and

selecting between use of the first traffic prediction model and the second traffic prediction model based on the first prediction error and the second prediction error.

3. The method of claim 2, wherein the calculating is based on a time and time horizon associated with the request.

4. The method of claim 1, wherein the determining comprises identifying the corresponding traffic data by identifying a subset of previously recorded traffic data that exhibits similar traffic conditions on a specific day of week, month or season that matches the associated day for the request.

5. The method of claim 1, comprising:

receiving information regarding an event that has one or more attributes that are correlated with reduction in traffic flow on one or more roads of a road network approaching the event;

calculating an influenced speed change and an influenced time shift, for a sensor associated with the road network, based on the information regarding the event; and

using the influenced speed change and the influenced time shift in application of the first traffic prediction model.

6. The method of claim 5, wherein calculating the influenced speed change and the influenced time shift comprises calculating based on attributes for the event comprising (i) start time, (ii) location, (iii) direction, (iv) event type, and (v) affected lanes.

7. The method of claim 1, wherein the previously recorded traffic data comprises data derived from mobile sensor data.

8. The method of claim 7, comprising generating the derived data by performing operations comprising:

calculating speeds for multiple mobile sensors from mobile sensor data with respect to connected road segments in a road network; and

generating a speed for a road segment of the connected road segments by calculating an aggregation of all speeds calculated for mobile sensors passing the road segment at a given time.

9. The method of claim 8, wherein the mobile sensor data is obtained from public transit vehicles.

10. A system comprising:

a user interface device; and

one or more computers operable to interact with the user interface device, the one or more computers comprising at least one processor and at least one memory device, and the one or more computers configured and arranged to perform operations comprising (i) receiving a request relating to traffic prediction, the request having an associated day and an associated time, (ii) determining how much to apply each of a first traffic prediction model and a second traffic prediction model based on previously recorded traffic data corresponding to the associated day and the associated time, wherein the first traffic prediction model comprises a moving average model that exhibits increased prediction accuracy as a prediction time horizon is reduced, and the second traffic prediction model comprises a historical average model that exhibits similar prediction accuracy across multiple prediction time horizons, and (iii) applying the first and second traffic prediction models in accordance with the determining to generate an output for use in relation to traffic prediction.

11. The system of claim 10, wherein the one or more computers are configured and arranged to perform operations comprising (i) receiving information regarding an event that has one or more attributes that are correlated with reduction in traffic flow on one or more roads of a road network approaching the event, (ii) calculating an influenced speed change and an influenced time shift, for a sensor associated with the road network, based on the information regarding the event including start time, location, direction, and severity of the event as compared with similar historical events, and (iii) using the influenced speed change and the influenced time shift in application of the first traffic prediction model.

12. The system of claim 10, wherein the one or more computers comprise a server operable to interact with the user interface device through a data communication network, and the user interface device is operable to interact with the server as a client.

13. The system of claim 12, wherein the user interface device comprises a mobile phone.

14. The system of claim 10, wherein the one or more computers comprises one personal computer, and the personal computer comprises the user interface device.

15. A method of predicting traffic on a road network in view of an event having an identified time and an identified location in the road network, the method comprising:

retrieving attributes from past events on the road network;

selecting a subset of the attributes that are correlated with traffic parameters comprising delayed traffic speeds, affected backlogs of vehicles, and amounts of time needed to clear backlogs of vehicles;

discovering corresponding values for the traffic parameters under all combinations of the selected attributes;

matching current attributes for the event in the road network to previous event attributes using the corresponding values for the traffic parameters to identify a subset of the past events; and

using the identified time, the identified location, and the subset of the past events to predict (i) a delayed traffic speed for the event, (ii) an affected backlog of vehicles on one or more roads approaching the event in the road network, and (iii) an amount of time needed for the affected backlog of vehicles to be cleared in the road network.

16. The method of claim 15, wherein the selected attributes comprise (i) start time, (ii) location, (iii) direction, (iv) event type, and (v) affected lanes.

17. The method of claim 15, wherein the past events comprise accidents, vehicle breakdowns, scheduled or unscheduled road closures or construction, emergencies, and social events, including concert and sporting events.

18. The method of claim 15, comprising predicting traffic on the road network using previously recorded traffic data comprising data derived from mobile sensor data.

19. The method of claim 18, comprising generating the derived data by performing operations comprising:

calculating speeds for multiple mobile sensors from mobile sensor data with respect to connected road segments in a road network; and

generating a speed for a road segment of the connected road segments by calculating an aggregation of all speeds calculated for mobile sensors passing the road segment at a given time.

20. The method of claim 19, wherein the mobile sensor data is obtained from public transit vehicles.