EP2553672A1

EP2553672A1 - Method, system and node for journey time measurement in a road network

Info

Publication number: EP2553672A1
Application number: EP10848656A
Authority: EP
Inventors: Dan Yu; Leiming Xu; Wei Qiu; Michael Sax
Original assignee: Siemens AG
Current assignee: Siemens AG
Priority date: 2010-03-31
Filing date: 2010-03-31
Publication date: 2013-02-06
Also published as: EP2553672A4; WO2011120194A1

Abstract

This invention discloses a method, system and node for journey time measurement in a road network. The method includes: designating multiple measured points in the road network, placing a node at each of the multiple measured points, and detecting characteristic of a car sequence sequentially passing through the measured point of the node; determining, at a first node, one or more neighbor nodes of the first node, and reporting to a neighbor node the characteristic of the car sequence of the first node, wherein the first node is any node in the road network; comparing, at the neighbor node, the characteristic of the car sequence of the first node and the characteristic of the car sequence of the neighbor node to find a matching position; obtaining, at the neighbor node, a journey time from the first node to the neighbor node according to the matching position.

Description

Method, System and Node for Journey Time Measurement in a Road Network

Field of the Invention

This invention relates to traffic information acquisition technique, and more particularly relates to a method, system and node for journey time measurement in a road network.

Background of the Invention

Journey time measurement system (JTMS) is a basic traffic information acquisition system in Information Transfer System (ITS). Journey time of link represents average traffic speed and status of traffic congestion on a road segment (i.e. a link) of a road network. The journey time of link provides foundation of traffic control at a local area. After collecting journey time of every link in the road network, journey time along a route (including a number of links) can be estimated by adding up all link journey times along this route. The journey time along a route provides evidence for route planning and optimization of traffic control in the whole urban area. In a JTMS, link coverage and flexibility are the most important problems to be solved.

Different methods are provided in the conventional art for journey time measurement. In these methods, a kind of detectors will be mounted at different positions (measured points) in the road network to detect some information of vehicles, and information from different detectors will be synthesized to estimate the journey time between detectors. To guarantee the system coverage and accuracy of journey time estimation, a lot of detectors will be deployed. Thus, a big volume of data will be generated from detectors and need to be processed. This results in challenges to communication capacity, computation capacity and cost of the system.

Methods of journey time measurement in the conventional art can be divided into two classes:

1. Individual vehicle recognition.

Systems, like automatic number plate recognition (ANPR) or RFID based JTMS, detect some unique identity of a vehicle and recognize the vehicle by its unique identity. For example, in an ANPR based JTMS shown in Figure 1 , cameras are installed at different sites to capture images of vehicles. Texts on license plates of vehicles will be recognized. By matching license plates from different sites, the system can identify a same vehicle at different sites and learn the time difference, i.e. the journey time, between the two sites. Another similar scheme is Dacolian's JTMS system, in which some unique feature (or characteristics) of a vehicle is extracted from images and coded as a signature. By comparing signatures, the system can identify same vehicles and learn the journey time.

This kind of method apparently works. However, it also raises privacy concern and encountered protest and resistance from privacy-aware people and organizations, as it would be very easy to collect the information about a specific vehicle at multiple points and to restructure the trace of the vehicle without too much effort.

Further, this method usually involves costly equipments and high computation complexity of algorithm. For example, in ANPR based or Dacolian's JTMS, cameras of high resolution and high frame rate must be used to make the video clear enough for OCR or other unique feature recognition. This also results in limited coverage area of a single camera.

2. Vehicle group matching.

Different to the method of individual vehicle recognition, the basic idea of vehicle group matching is not to' r'ecogriize an individual vehicle, but to match a vehicle group (also called as a car sequence or a vehicle sequence) at different places. Usually, vehicles in the same road segment at the same time will remain a similar speed. Thus, if two vehicles are close to each other when they arrive at the beginning of the road segment, their relative locations will not change very much at the end of this road segment, especially when the road segment is not so long. As shown in Figure 2, a first feature sequence (i.e. some features of a first sequence of cars) is detected at the beginning and a second feature sequence (features of a second sequence of cars) is detected at the end. Then, the first and second feature sequences are reported to some device and compared by the device to find whether the car sequences at the two places are matched. If the first and second car sequences are matched, the journey time through this road segment can be obtained. Specifically, features detected in such a system do not have to differentiate this vehicle from others. Any common features, such as vehicle color, shape, weight etc., can be utilized for vehicle group matching. For example, a solution is proposed to use a loop to detect the length of a vehicle and use characteristics in length sequence to recognize a vehicle sequence.

It is clear that the method of vehicle group matching does not recognize any identity of a vehicle. Thus, first of all, privacy is protected. Second, since the features detected are simpler than those in the method of individual vehicle recognition, detector used in this method will■^■ be simpler and cheaper. And, communication bandwidth requirement may be lower than ANPR based JTMS because, in this method, only features are transmitted from detectors to a matching device, such as a server, while in many implementations of ANPR based JTMS, raw images are required to be transmitted to a back-end ANPR server. However, this method has its shortcomings, for example, performance will degrade as distance between two detected sites increases.

With these methods, currently, there are two architectures to construct a JTMS.

A. Separate processing.

A system is adapted to measure only one link (a road segment between 2 intersections) or one route (including consecutive links), wherein a detector is deployed at each end of the link or route. For every link or route to measure, two new detectors are deployed (at two ends respectively). Detectors and data are not shared between these separate JTMSs.

B. Centralized processing.

All detectors are connected to a network, and send their data to a Traffic Management Center (TMC). The centralized TMC processes these data, and decides between which two detectors data should be compared and journey time should be calculated. This architecture is more flexible than architecture A, but need more communication capacity and computation power at the TMC.

In view of the above, a better method and system for journey time measurement in a road network are required. Summary of the Invention

In the present invention, a new method, system architecture of JTMS (i.e. distributed JTMS) and a node structure are provided, in which the system can be constructed automatically and configured flexibly and data processing can be distributed to every node, so that the problems of computation capacity, communication capacity and system flexibility can be solved.

The technical solution of the present invention is as follows.

A method for journey time measurement in a road network includes:

designating multiple measured points in the road network, placing a node at each of the multiple measured points, and detecting characteristic of a car sequence sequentially passing through the measured point of the node;

determining, at a first node, one or more neighbor nodes of the first node, and reporting to a neighbor node the characteristic of the car sequence of the first node, wherein the first node is any node in the road network;

comparing, at the neighbor node, the characteristic of the car sequence of the first node and the characteristic of the car sequence of the neighbor node to find a matching position;

obtaining, at the neighbor node, a journey time from the first node to the neighbor node according to the matching position.

Optionally, detecting characteristic of a car sequence sequentially passing through the measured point of the node includes:

obtaining variation of magnetic field caused by each of the cars sequentially passing through the measured point of the node; or, obtaining color, shape or voiceprint of each of the cars; or, obtaining any combination of variation of magnetic field, color, shape and voiceprint of each of the cars.

Optionally, determining, at a first node, one or more neighbor nodes of the first node includes:

broadcasting, by the first node, the characteristic of the car sequence of the first node during a period of time to part of or all the other nodes in the road network; any other node performing characteristic matching between the characteristic of the car sequence detected by itself and the characteristic of the car sequence of the first node, and sending a neighborhood indication to the first node when the two characteristic are found to be matched;

recording, by the first node, any other node that has sent the neighborhood indication to the first node as a neighbor node.

configuring for the first node one or more other nodes as its neighbor nodes according to the map of the road network, and storing the configuration on the first node.

Optionally, comparing, at the neighbor node, the characteristic of the car sequence of the first node and the characteristic of the car sequence of the neighbor node to find a matching position includes:

selecting a first segment from the characteristic of the car sequence of the first node;

selecting multiple second segments from the characteristic of the car sequence of the neighbor node, wherein each of the multiple second segments has the same length as the first segment; and

performing correlation between the first segment and each of the multiple second segments, and determining the matching position according to the maximum correlation coefficient.

Optionally, obtaining, at the neighbor node, a journey time from the first node to the neighbor node according to the matching position includes:

determining a first time of the matching position on the characteristic of the car sequence of the first node;

determining a second time of the matching position on the characteristic of the car sequence of the neighbor node; and

obtaining the delay between the second time and the first time as the journey time from the first node to the neighbor node. A system for journey time measurement in a road network includes:

multiple nodes, each of which is placed at a measured point and adapted to detect characteristic of a car sequence sequentially passing through the measured point of the node, wherein multiple measured points are designated in the road network; and

a first node among the multiple nodes is further adapted to determine one or more neighbor nodes of the first node, and report to the neighbor nodes the characteristic of the car sequence of the first node;

any of the neighbor nodes is adapted to compare the characteristic of the car sequence of the first node and the characteristic of the car sequence of the neighbor node to find a matching position, and obtain a journey time from the first node to the neighbor node according to the matching position.

Optionally, the first node is adapted to broadcast the characteristic of the car sequence of the first node during a period of time to part of or all the other nodes in the road network;

any other node is adapted to perform characteristic matching between the characteristic of the car sequence detected by itself and the characteristic of the car sequence of the first node, and send a neighborhood indication to the first node when the two characteristic are found to be matched; and

the first node is adapted to record any other node that has sent the neighborhood indication to the first node as a neighbor node.

The system further includes:

a management center, adapted to provide configuration information to the first node for informing the first node of its neighbor nodes, and receive the journey time from the first node to any of its neighbor nodes reported by the neighbor node.

A node placed at a measured point in a road network for journey time measurement includes:

a detector, adapted to detect characteristic of a car sequence sequentially passing through the measured point of its node; a communication unit, adapted to receive characteristic of a car sequence of another node and provide the characteristic of the car sequence of another node to a computation unit; and

the computation unit, adapted to compare me characteristic of the car sequence of another node and the characteristic of the car sequence of its node to find a matching position, and obtain a journey time from another node to its node according to the matching position.

The node further includes:

a neighborhood management unit, adapted to record any other node that has sent a neighborhood indication to its node into a neighbor list, and instruct the communication unit to send the characteristic of the car sequence of its node to neighbor nodes in the neighbor list; and

the communication unit is further adapted to broadcast the characteristic of the car sequence of its node detected by the detector during a period of time to part of or all the other nodes in the road network, and report the characteristic of the car sequence of its node to the neighbor nodes after receiving the instruction from the neighborhood management unit.

The node further includes:

a neighborhood management unit, adapted to update its neighbor list according to configuration information from a management center, and inform the communication unit to send the characteristic of the car sequence of its node to neighbor nodes in the neighbor list.

Brief Description of Drawings

FIG. 1 is an ANPR based JTMS in the conventional art;

FIG. 2 is a JTMS using vehicle group matching in the conventional art;

FIG. 3 is an example of a distributed JTMS provided in an embodiment of the present invention;

FIG. 4 is an automatic topology learning procedure in an embodiment of the present invention; FIG. 5 is a structure of a node used in a distributed JTMS according to an embodiment of the present invention.

Detailed Description of the Invention

This invention is hereinafter further described in detail with reference to the accompanying drawings as well as embodiments so as to make the objective, technical solution and merits thereof more apparent.

In an embodiment of the present invention, a node including a detector (also named as a sensor) will be mounted at every measured point in a road network, and all the nodes will be connected to form a JTMS. It should be noted that multiple measured points are designated in the road network according to characteristics of real roads, to guarantee that the actual road distance of two measured points is not so long.

The detector can be any kind of detectors capable of obtaining characteristic of a car sequence, such as obtaining variation of magnetic field caused by each of the cars sequentially passing through the measured point of the node; or obtaining color, shape or voiceprint of each of the cars; or obtaining any combination of variation of magnetic field, color, shape and voiceprint of each of the cars.

Then, data will be transmitted between nearby nodes, and journey time of every link will be calculated at an adjacent node and only results will be reported to a management center (such as a TMC).

For example, in an illustration shown in Figure 3, nodes 1-8 are located at some points (referred to as measured points) in a street map of the road network, wherein the arrow represents a road direction.

(1) Nodes 1 and 7 send their data to node 2 since node 2 is their neighbor node.

(2) Node 2 finds matches between data detected by itself and data from nodes 1 and 7, and knows journey time of link between nodes 1 and 2 and journey time of link between nodes 7 and 2.

Similarly, node 4 will learn journey time of 3 <— > 4; node 5 will learn journey time of 4 <-- 5; node 6 will learn journey time of 3 <— 6 and 4 6; node 8 will learn journey time of 7 ^--> 8 etc.

Specifically, taking journey time of link between nodes 1 and 2 as an example, the procedure of journey time calculation is as follows: Step 1 : Node 2 selects a first segment from characteristic of the car sequence of node 1 , and selects multiple second segments from characteristic of the car sequence of node 2, wherein each of the multiple second segments has the same length as the first segment.

Step 2: Node 2 performs correlation between the first segment and each of the multiple second segments, and determines a matching position according to the maximum correlation coefficient.

Step 3 : Node 2 determines a first time of the matching position on the characteristic of the car sequence of node 1, determines a second time of the matching position on the characteristic of the car sequence of node 2, and obtains the delay between the second time and the first time as the journey time from node 1 to node 2.

In the distributed JTMS, every node knows its neighbor nodes, and will send data generated by itself to the neighbor nodes. Also, every node is possible to serve as a neighbor node of other nodes, will receive data from the other nodes and is responsible to calculate the journey times between the other nodes and itself.

In every node, knowledge of neighborhood can be configured by a TMC according to a map of the road network, provided that the TMC knows location of every node and knows which links' journey times are wanted. That is, the TMC can tell every node to which nodes it should send data. Then, the node will configure one or more other nodes as its neighbor nodes according to the instruction from the TMC, and store the configuration on itself.

Or, neighborhood relationship can be learnt by nodes automatically. First, when the system starts up, every node will broadcast its data, i.e. characteristic of a car sequence passing through itself, to other nodes in the system. By searching matches between data from itself and data received from other nodes, a node can learn which nodes have direct links to it and feed back matching information to the other nodes having direct links. Then, every node will set the other nodes providing matching information to it as its neighbor nodes. At the second phase, all nodes stop broadcasting and only send data to their neighbor nodes.

A possible automatic topology; learning procedure can be as illustrated in Figure

4.

(1) Node A performs feature detection and broadcasts the feature detected to nodes B-D and nodes U- W. At the same time, nodes B-D and nodes U-W perform feature detection at their own points, respectively.

(2) Each of nodes B-D and nodes U-W compares the feature detected by itself and the feature broadcasted by node A for feature sequence matching.

(3) It is supposed that matches are found in nodes B-D. Then, nodes B-D send a neighborhood indication to node A.

(4) Node A records nodes B-D as its neighbor nodes, and only transmits feature of its car sequence to its neighbor nodes. As to nodes U-W, since no matches have been found, node A will not provide feature of its car sequence to nodes U-W any longer.

Further, it should be pointed out that even if node B is a neighbor node of node A, node A may not necessarily be a neighbor node of node B. For example, in Figure 3, node 3 is a neighbor node of node 2, while node 2 is not a neighbor node of node 3.

Accordingly, the structure of a node used in the distributed JTMS is shown in Figure 5. The node includes: a detector, a communication unit, a computation unit and a neighborhood management unit.

The detector is adapted for detecting features of vehicles on road. Any features, which are detectable and can be regarded as unchanged between two detecting points, can be utilized by the detector. Specifically, the features can be such as color, shape parameters, weight, voiceprint and even gray level of image. Of course, identities, such as RFID or license plate number, also can be used here. Detected features may be sent to the computation unit for searching matches with data from other nodes, and/or sent to neighbor nodes through the communication unit.

The communication unit is adapted to receive configuration information from a management center, report journey time measurement results to the management center, and receive/send data from/to other nodes.

The computation unit is adapted to find matching vehicle groups from data of itself and data of other nodes it received, and calculate journey times. Usually, the computation unit needs a memory to buffer a certain time period of data from other nodes and from itself. In case of some identity used, it will retrogress to individual vehicle matching. The neighborhood management unit is adapted for maintaining a list of neighbor nodes, and informing the communication unit to whom data should be sent.

In automatic mode, at first, the neighborhood management unit regards all other nodes as its neighbor nodes, and the communication unit will broadcast data to all the other nodes. After neighborhood indications are received by the communication unit, the neighborhood management unit will extract neighborhood information from the communication unit, and update its list.

In manual mode, the neighborhood management unit is adapted for updating its neighbor list according to configuration information from the management center.

To sum up, in the above mentioned scheme of the present invention:

a) Calculation is distributed to nodes, and computation power bottleneck is prevented.

b) Data are only transmitted to nearby nodes. Thus, communications are always limited in a local range. No big volume of data flow will be involved in the system. c) Neighborhood can be discovered by nodes automatically. Thus, the system structure can be constructed by the nodes themselves. In this way, system deployment becomes easy, and work of pre-configuration can be saved.

d) Measured links can be selected easily by configuration, and nodes are not dedicated to any measured link or route.

The foregoing is only preferred embodiments of the present invention and is not for use in limiting the present invention. Any modification, equivalent replacement or improvement made under the spirit and principles of the present invention is included in the protection scope thereof.

Claims

1. A method for journey time measurement in a road network, comprising:

2. The method according to claim 1 , wherein detecting characteristic of a car sequence sequentially passing through the measured point of the node comprises: obtaining variation of magnetic field caused by each of the cars sequentially passing through the measured point of the node; or, obtaining color, shape or voiceprint of each of the cars; or, obtaining any combination of variation of magnetic field, color, shape and voiceprint of each of the cars.

3. The method according to claim 1 , wherein determining, at a first node, one or more neighbor nodes of the first node comprises:

4. The method according to claim 1, wherein determining, at a first node, one or more neighbor nodes of the first node comprises:

5. The method according to any of claims 1-4, wherein comparing, at the neighbor node, the characteristic of the car sequence of the first node and the characteristic of the car sequence of the neighbor node to find a matching position comprises:

6. The method according to any of claims 1 -4, wherein obtaining, at the neighbor node, a journey time from the first node to the neighbor node according to the matching position comprises:

determining a first time of the matching position on the characteristic of the car sequence of the first node; '

obtaining the delay between the second time and the first time as the journey time from the first node to the neighbor node.

7. A system for journey time measurement in a road network, comprising:

multiple nodes, each of which is placed at a measured point and adapted to detect characteristic of a car sequence sequentially passing through the measured point of the node, wherein multiple measured points are designated in the road network; and a first node among the multiple nodes is further adapted to determine one or more neighbor nodes of the first node, and report to the neighbor nodes the characteristic of the car sequence of the first node;

8. The system according to claim 7, wherein the first node is adapted to broadcast the characteristic of the car sequence of the first node during a period of time to part of or all the other nodes in the road network;

any other node is adapted to perform characteristic matching between the characteristic of the car sequence detected by itself and the characteristic of the car sequence of the first node, and send fa neighborhood indication to the first node when the two characteristic are found to be matched; and

9. The system according to claim 7, further comprising:

10. A node placed at a measured point in a road network for journey time measurement, comprising:

a detector, adapted to detect characteristic of a car sequence sequentially passing through the measured point of its node;

a communication unit, adapted to receive characteristic of a car sequence of another node and provide the characteristic of the car sequence of another node to a computation unit; and

the computation unit, adapted to compare the characteristic of the car sequence of another node and the characteristic of the car sequence of its node to find a matching position, and obtain a journey time from another node to its node according to the matching position.

1 1. The node according to claim 10, further comprising:

12. The node according to claim 10, further comprising: