P/00/01 1 Regulation 3.2 AUSTRALIA Patents Act 1990 ORIGINAL COMPLETE SPECIFICATION STANDARD PATENT Invention Title: Method for Dynamically Integrating Traffic Data Applicant: Intelematics Australia Pty Ltd The following statement is a full description of this invention, including the best method of performing it known to me: 1 2 METHOD FOR DYNAMICALLY INTEGRATING TRAFFIC DATA FIELD OF THE INVENTION The present invention relates to the field of traffic information services. In particular 5 the present invention relates to a method of dynamically integrating traffic data from traffic probes and from road sensors. The method may include acquiring, processing and/or producing traffic data. BACKGROUND OF THE INVENTION 10 A number of techniques may be used to measure, predict and/or estimate average travel speeds including travel times and/or delays on road networks. The techniques may generate traffic data for operating traffic information services and for systems that require an estimate of current or future speeds including travel times and/or delays on road segments. 15 Prior art methods for producing traffic data typically fall into two groups. One method group includes use of traffic models coupled with dynamic information or data obtained from sensor networks (sensor-model method). Another method group includes use of traffic probes wherein speed of vehicles (including individuals in the ?0 vehicles) is remotely measured or estimated in real time or near real time using a range of technologies including GPS and/or cellular network triangulation (probe method). However a disadvantage of traffic models based on data from sensor networks alone 25 is that they are inaccurate under certain conditions. This is due in part to characteristics of individual road sensors. As a result prior art traffic models using standard calibration parameters produce relatively crude traffic congestion data. SUMMARY OF THE INVENTION 30 The method of the present invention includes a methodology that involves a combination or integration of previously independent techniques for acquiring traffic data. The particular methodology may provide a superior result when compared to independent approaches taken alone. The methodology of the present invention is Spec as filed - 872145 3 distinct from a process of simply fusing speed information generated from multiple techniques into a combined or weighted average. The method of the present invention may normalize data produced by a standard 5 traffic model by comparing it to data from traffic probes. Assuming that a comparison shows a statistically significant difference between data from the standard model and data from the traffic probes, the standard model data may be normalized. The standard model data may be normalized by applying normalizing coefficients to raw sensor data provided by the standard model. The normalizing coefficients may take 10 into account delays to vehicles along a segment of road between two points or nodes (segment movement delay) and/or delays to vehicles at an intersection of two or more roads (road intersection delay). The normalizing coefficients may make use of probe data from a starting point of a time period associated with a calibration sequence. The normalized calibration parameters may be fine tuned over a period of time to take 15 into account average length of road segments along which speeds are calculated and average intersection characteristics such as number of through lanes, turn lanes, turning movement restrictions, etc. Data from traffic probes may be collected and recorded at a single point in time. The ?0 data may be stamped with speed, bearing, positional quality and geographical coordinates either as longitude/latitude or as X-Y pairs. Each record of probe data may be matched to geometry of available road segments with a valid road sensor data link. Multiple matches for a single probe data record may be ranked with significance to a closest matched road segment and with an appropriate bearing 25 value when compared to a linear direction or angle of the road segment. The resulting matched probe data may represent a selection of most probable traffic ground truth data from probe vehicles along a respective road segment. Data that has been identified as compromised may be excluded from the matched 30 probe data. The latter may include suspected long periods of no movement which is not likely to be caused by traffic congestion, records with poor GPS positional quality, frequent records with speeds greater than speed limits on the associated road segments, etc. Spec as fied - 872145 4 The matched probe data may be tested for statistical difference when compared to traffic congestion indicators produced by a traffic information service. The testing may include a two-sample t-test. A two-sample t-test may be used for testing a hypothesis about the location two data sets being equal. If the two data sets are 5 statistically different, calibration parameters may be recalculated using the matched probe data set as a base. The recalculated parameters may be used in the traffic model until the next cycle when another recalculation may be performed. Through the above methodology, a relatively small number of traffic probes may be 10 usefully leveraged when compared with methods that use traffic probes as a source of real-time information only. Dynamic integration of traffic probes data and road sensors data is superior to traditional average based and/or weighted summary. The traffic probe data may be 15 used to dynamically calibrate traffic information models and/or systems in near real time. The traffic information model and/or system may be parameterised and designed to accept dynamic adjustment and decision-making output based on specified near real-time processes. Statistical techniques may be used to acquire, validate, calculate and produce necessary data for traffic model adjustment. The ?0 resulting adjusted traffic model may be capable of validating its accuracy of fit using specific statistical techniques and regular sanity checks and comparison to traffic trends data. According to the present invention there is provided a method of dynamically 25 integrating data from traffic probes and from road sensors, said method including: measuring speeds of said traffic probes over defined road segments; uploading said measured probe speeds to a processing system; comparing said measured probe speeds to speeds predicted by said road sensors to produce variance data; and 30 recalibrating said predicted speeds at least periodically based on said variance data. The step of comparing may make use of at least one statistical technique. The step of recalibrating may make use of at least one statistical technique. The or each Spec as filed - 872145 5 statistical technique may be used to acquire, validate and/or calculate traffic information to provide the variance data. The present invention may provide a dynamically adjustable traffic information service 5 using data from probe vehicles to enhance quality and accuracy of data available from road sensors. Integrating data from probe vehicles in real-time may provide a valuable opportunity for enhancing and improving quality over otherwise static calibration techniques. Quality of traffic information services may be improved by integrating data from probe vehicles in the setting and by validation and usage of 10 traffic model parameters in real-time (or near real-time). By dynamically updating traffic model parameters using current real-time traffic conditions, traffic information services may provide motorists with relatively more accurate and tangible information on traffic conditions. It may also be more 15 responsive to irregular traffic trends on road segments, road sensor faults and propagation of grid locks. The method of the present invention may expand use of data from probe vehicles including a dynamic application to calibrating a traffic modelling process. The method ?0 of the present invention may make use of statistically proven and quality enhanced calibration model parameters. DESCRIPTION OF A PREFERRED EMBODIMENT A preferred embodiment of the present invention will now be described with reference 25 to the accompanying drawing wherein: Figure 1 shows a flow diagram of a method for integrating data from traffic probes with data obtained from road sensors. 30 Referring to Figure 1 conventional traffic congestion indicators are used to produce dynamic traffic congestion information from road sensor networks including loop data associated with a traffic service system 10. A traffic congestion calculation is performed at step 11 using a standard traffic model including standard calibration Sped as Wed - 872145 6 parameters. The traffic congestion indicators are also used to produce a list of road segments with available probe data for a corresponding period. Following probabilistic matching at step 12 to probe locations the data produced by 5 the standard traffic model is compared at step 13 to the probe data. Assuming that the comparison shows a statistically significant difference, the standard calibration model (step 11) is updated or normalized at step 14 by applying normalizing coefficients to the data provided by the standard model. 10 Assuming that the comparison shows a statistically significant difference in a subsequent cycle the normalized model is compared to the probe data and another recalculation may be performed. Assuming that the result of the comparison is acceptable, i.e. the comparison does 15 not show a statistically significant difference, the integrated traffic information is published at step 15 to a traffic information service. In a preferred embodiment data from sensor networks is integrated with data from traffic probes by a methodology that includes at least the following steps: ?0 1. produce statistical confidence level of matching probe data to use for a calibration model: (i) exclude outliers from long periods of no movement; (ii) exclude records with poor GPS positional quality; (iii) calculate average speed for trips on road segments; 25 2. for road segments where reliable probe data is available, calculate similarity/difference tests; 3. if a difference between traffic congestion indicators and selected probe data is found to be statistically significant, update the calibration model with new values from probe dataset; 30 4. otherwise, accept traffic congestion indicators from standard calibration model. Fields of application for the method of the present invention may include: - Traffic Analysis - Traffic Prediction Speci as Ned - 872145 7 - Spatial Analysis and Integration with Traffic Information Services - GPS Route Calculation and Journey Statistics - Road Travel Time calculation - Road Speed Calculation 5 - Device Monitoring and Tracking Data Manipulation - Database Systems. Finally, it is to be understood that various alterations, modifications and/or additions may be introduced into the constructions and arrangements of parts previously 10 described without departing from the spirit or ambit of the invention. Spec as filed - 872145