EP2374117B1 - Vehicle tracking system, vehicle infrastructure provided with vehicle tracking system and method for tracking vehicles - Google Patents
Vehicle tracking system, vehicle infrastructure provided with vehicle tracking system and method for tracking vehicles Download PDFInfo
- Publication number
- EP2374117B1 EP2374117B1 EP09771423.2A EP09771423A EP2374117B1 EP 2374117 B1 EP2374117 B1 EP 2374117B1 EP 09771423 A EP09771423 A EP 09771423A EP 2374117 B1 EP2374117 B1 EP 2374117B1
- Authority
- EP
- European Patent Office
- Prior art keywords
- vehicle
- message
- facility
- sensor nodes
- tracking system
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims description 101
- 238000001514 detection method Methods 0.000 claims description 119
- 230000001360 synchronised effect Effects 0.000 claims description 13
- 238000004891 communication Methods 0.000 claims description 12
- 230000005540 biological transmission Effects 0.000 claims description 10
- 238000012544 monitoring process Methods 0.000 claims description 4
- 239000013598 vector Substances 0.000 description 76
- 238000005070 sampling Methods 0.000 description 66
- 230000006870 function Effects 0.000 description 34
- 239000011159 matrix material Substances 0.000 description 27
- 238000005259 measurement Methods 0.000 description 25
- 238000012545 processing Methods 0.000 description 22
- 238000004422 calculation algorithm Methods 0.000 description 10
- 230000008569 process Effects 0.000 description 8
- 238000004458 analytical method Methods 0.000 description 7
- 238000013459 approach Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 6
- 230000033001 locomotion Effects 0.000 description 6
- 230000005291 magnetic effect Effects 0.000 description 6
- 230000015654 memory Effects 0.000 description 6
- 230000001133 acceleration Effects 0.000 description 5
- 230000001419 dependent effect Effects 0.000 description 5
- 238000013461 design Methods 0.000 description 5
- 238000009472 formulation Methods 0.000 description 4
- 239000000203 mixture Substances 0.000 description 4
- 238000012546 transfer Methods 0.000 description 4
- 230000006399 behavior Effects 0.000 description 3
- 230000006698 induction Effects 0.000 description 3
- 239000002245 particle Substances 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 2
- 238000009795 derivation Methods 0.000 description 2
- 238000005192 partition Methods 0.000 description 2
- 230000011664 signaling Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 239000013256 coordination polymer Substances 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005183 dynamical system Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 239000003302 ferromagnetic material Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000012010 growth Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 229910052751 metal Inorganic materials 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
Definitions
- Vehicle tracking system vehicle infrastructure provided with vehicle tracking system and method for tracking vehicles.
- the present invention relates to a vehicle tracking system.
- the present invention relates to a vehicle infrastructure provided with a vehicle tracking system.
- the present invention further relates to a method for tracking vehicles.
- PeMS traffic performance measurement system
- US5801943 describes a wide area surveillance system for application to large road networks.
- the system employs smart sensors to identify plural individual vehicles in the network. These vehicles are tracked on an individual basis, and the system derives the behavior of the vehicle. Furthermore, the system derives traffic behavior on a local basis, across roadway links, and in sections of the network. Processing in the system is divided into multiple processing layers, with geographical separation of tasks.
- the vehicle tracking system according to the present invention comprises
- the vehicle tracking system in the vehicle tracking system according to the present invention vehicles can be tracked with relatively simple and cheap means. Smart sensors are not necessary. It is sufficient that the sensor nodes sense an occupancy state, i.e. whether a detection area associated with the sensor node is occupied by a vehicle or not and that they merely provide a message that indicates whether the occupancy state is changed.
- the relatively cheap and simple construction of the sensor nodes contributes to an economically feasible application in vehicle tracking systems for large vehicle infrastructures.
- the message may additionally include the value of the occupancy state after the change was detected.
- the plurality of sensor nodes arranged in the vehicle infrastructure having at least the above-mentioned density provide a course image of the vehicles present at the vehicle infrastructure. However, as compared to an image provided by cameras, the image provided by the plurality of sensor nodes is always captured from the same perspective. This facilitates further processing.
- the association facility selects for which state information the received messages are relevant, and provides the selected messages to the state updating facility. In this way the state updating facility can operate more efficiently, than in case no selection takes place.
- the present invention is in particular suitable for tracking vehicles.
- suitable sensor elements to be used in the sensor nodes are for example magneto restrictive sensors. These sensors determine whether their associated detection area is occupied by detection of a perturbation of the earth magnetic field.
- magnetic loop sensors may be used, which detect a change of inductance caused by the presence of ferromagnetic material.
- each sensor node is provided with a wireless transmission facility that transmits the preprocessed data, e.g. the occupance status or an indication of a change thereof to a data to a receiver facility coupled to the message interpreter.
- a wireless transmission facility that transmits the preprocessed data, e.g. the occupance status or an indication of a change thereof to a data to a receiver facility coupled to the message interpreter.
- the absence of wiring towards the message interpreter makes the installation easier and cost effective.
- the sensor nodes provide their message at an event basis, e.g. if a perturbation of the earth magnetic field exceeds a threshold value. This reduces communication load of the message interpreter and minimizes power consumption of the sensor nodes.
- the density of the sensor nodes are at least 0.6 per square meter. Due to the relatively high density of the sensor nodes in the traffic infrastructure in said embodiment an individual failure of a sensor or of an individual sensor does not have serious consequences for the estimation of the states of the traffic participants. Accordingly, any return transmission from the message interpreter to the sensor nodes to actively verify the occupancy state is superfluous, which is also favorable for a low power consumption of the sensor nodes.
- the vehicle tracking system may in addition to the plurality of sensor nodes arranged in the traffic infrastructure comprise one or more cameras.
- a camera may be used for example if a perturbation of the earth magnetic field can not be measured. This is the case for example if (parts of) the infra structure comprises metal components e.g. a bridge.
- the detection areas of the sensor elements are complementary. The detection areas may overlap, or spaces may exist between the detection areas, but it is required that the detection area of the sensor be smaller than the vehicles to be tracked.
- the sensor elements are point detectors.
- the sensor nodes can be either randomly distributed over the vehicle infrastructure or placed in a pattern optimized for the vehicle tracking problem in hand.
- the vehicle tracking system comprises a plurality of system modules, each module comprising a respective subset of the plurality of sensor nodes for monitoring a respective section of the vehicle infrastructure and a respective message interpreter, the vehicle tracking system has a communication facility for enabling system modules of mutually neighboring sections to exchange state and detection information. In this way the vehicle tracking system can be easily expanded if required.
- a new system module need only to communicate with the system modules arranged for neighboring sections. Neighboring sections may be arranged in one dimensional scheme, e.g. in case of a narrow road. For example if a certain road is already provided with an vehicle tracking system, it is sufficient to provide for a communication facility between the system module for the last section of said vehicle tracking system and the new system module for the appended section.
- the new module may communicate with other modules neighboring in various directions.
- the system modules merely need to exchange state information and vehicle-detection information (i.e. the unprocessed sensor signals) in a limited subarea of the respective sections, the amount of communication between the system modules is modest resulting in a scalable vehicle tracking system.
- the association facility associates the messages provided by the sensor nodes or neighboring system modules with the state information present in the vehicle data base facility. In other words the association facility determines the probability that the detections are caused by a particular vehicle for which state information is present in the vehicle data base facility. If the messages cannot be associated with state information of an already identified vehicle here or in the neighboring system module, a new entry may be added to the database. Alternatively, the entry for the new vehicle may be added by a separate procedure.
- the vehicle infrastructure may have an access with a vehicle identification facility that provides for an identification of every vehicle that enters the infrastructure.
- the individual sensor nodes do not need to provide other information than an occupancy status of their associated detection area.
- the sensor node may associate its own signal with a color, shape, or other signature of the tracked vehicles to facilitate or obviate association by the message interpreter.
- An association facility for associating the detection signals obtained on asynchronous basis with state information of a particular vehicle may be based on one of the following methods.
- Gating comprises forming a gate around the predicted measurement of a vehicle.
- the size and shape of the gate are chosen in such a way that unlikely messages are precluded to be associated with this particular vehicle-track.
- the method determines a statistical, quadratic distance d oi 2 from vehicle i.
- a measurement y is associated with the state-vector x oi of vehicle i if d oi 2 ⁇ G , with G some constant threshold and d oi 2 equal to:
- the Nearest Neighbor method also uses a gate, but it can handle overlapping gates.
- the sum of all possible combinations to associate a certain measurement to a certain track is analyzed.
- the chosen combination associates the most measurements to a track for a minimum sum of distances.
- JPDA Probabilistic Data Association
- a Multiple Hypothesis Tracker allows that the state-vector of a single vehicle can has multiple tracks.
- This method resembles to the Particle filter as described in B. Ristic, S. arulampalam, and N. Gordon, "Beyond the Kalman filter: Particle filter for tracking applications", 2002 . Therein, each state is estimated by simulating N states with each a different probability. It is a drawback of this method that it requires a high computational power.
- a further data association method is the Markov chain Monte Carlo data association (MCMCDA). All observations are used to classify and cluster them. To that end the whole set of observations is divided into a number of partitions represented by the set w. This is done n mc times resulting in n mc sets of w, i.e. possible partitions. The set of w with the highest probability, given the number of vehicles in the previous sample instant, is chosen and the state-vectors of the tracks a are updated according the partitioned observation. The computational time can be decreased by not using the total history of observations, but by using a moving horizon.
- a downside of this method is that each observation can belong to at most one vehicle and, making this method unsuitable for event-based state-estimation.
- step of associating may comprise
- the state of a vehicle can also be estimated at a point in time later than the last message, but before a new message has arrived. In that case the error covariance matrix is bounded, as it is known that the state change of the vehicle must be within the detection boundaries of the sensor node.
- first, second, third etc. may be used herein to describe various elements, components, and/or sections, these elements, components, and/or sections should not be limited by these terms. These terms are only used to distinguish one element, component or section from another element, component, and/or section. Thus, a first element, component, and/or section discussed below could be termed a second element, component, and/or section without departing from the teachings of the present invention.
- Figure 1 and 2 show a first and a second view of an embodiment of a vehicle infrastructure 80 provided with a vehicle tracking system.
- the vehicle infrastructure is intended to allow stationary and/or moving vehicles 70 thereon, e.g. a road or a parking place.
- the vehicle infrastructure may be part of a public or private space, e.g. a recreational park.
- the vehicle tracking system comprises a plurality of sensor nodes 10 that each provide a message indicative for an occupancy status of a detection area of the vehicle infrastructure monitored by said sensor node 10. As shown therein the sensor nodes are randomly distributed over the vehicle infrastructure.
- the vehicle tracking system comprises a message interpretator MI, each comprising a vehicle database facility, an association facility and a state updating facility.
- Each message interpretator is responsible for handling messages D from a respective section 80A, 80B, 80C, 80D of the vehicle infrastructure 80.
- Figure 3 is another schematic view of the vehicle tracking system.
- Figure 3 shows how sensor nodes 10 transmit (detection) messages to a message interpreter MI in their neighborhood.
- the message interpreters MI may also communicate to each other via a communication channel 60 to indicate that a vehicle crosses a boundary between their respective sections and to exchange a status of such a vehicle.
- the vehicle tracking system comprises a plurality of system modules MD1, MD2, MD3. Although three modules are shown in this example, any number of system modules is possible, dependent on the application. For example for an isolated vehicle infra structure, e.g. an intersection of roads a single module may be applicable, while on a long road thousands of modules may be present.
- Each module MD 1, MD2, MD3 comprises a respective subset of the plurality of sensor nodes 10 for monitoring a respective section of the vehicle infrastructure and a respective message interpreter MI.
- the vehicle tracking system further has a communication facility 60 for enabling system modules MD1, MD2, MD3 of mutually neighboring sections to exchange state information.
- messages from the sensor nodes are directly transmitted to a message interpretor.
- the sensor nodes may form a network that routes the messages to the message interpreters. In that case the transmitters may have a short transmission range.
- Figure 4 schematically shows a part of the vehicle infrastructure that is provided with a plurality of sensor nodes j having position c j .
- the sensor nodes have a detection area with radius R.
- a vehicle i is present at the infrastructure having a position (v i x , v i y ). In this case if the vehicle substantially covers the detection area the sensor node indicates that the detection area is occupied as indicated in gray. Otherwise the sensor node indicates that the detection area is not occupied (white).
- the fraction of the detection area that should be covered before an occupied status is detected may deviate from the above-mentioned 50% depending on the type of vehicle.
- Figure 5 schematically illustrates the signal flow for the sensor node 10, having sensor element 12, a processing unit 14 (with memory), and a radio link 16.
- the sensor element 12 is capable of sensing the proximity of the vehicles to be tracked.
- the processing unit 14 determines if a vehicle is present or absent on the basis of the signals from the sensor element 12. If an occupancy status of the detection area of the sensor changes, the processing unit 14 initiates a transmission of a message D indicating the new occupancy status.
- the message may include a time stamp indicative of the time t at which the new occupancy status occurred.
- the sensor nodes may transmit occupancy status information on a periodical basis for example. However, an event-based transmission enables a lower power use.
- the message D sent should reach at least one message interpreter MI.
- the sensor element 12 is a magnetoresistive component, which measures the disturbance on the earth magnetic field induced by the vehicles.
- a magnetic rod or loop antenna may be used to detect the occupancy by a vehicle.
- FIG. 6 shows a possible implementation of the hardware involved for the sensor node 10 of Figure 5 .
- the sensor element 12 is coupled via an A/D converter 13 to a microcontroller 14 that has access to a memory 15, and that further controls a radio transmitter 16 coupled to an antenna 17.
- Figure 7 schematically shows a method performed by a sensor node to generate a message indicative for occupancy status of a detection area of the sensor node.
- Step S1 initialization
- Step S2 input from the A/D converter
- Step S3 offset is removed from the sensed value.
- step S4 it is determined whether the occupancy state of the detection area as reported by the last message transmitted by the sensor node was ON (vehicle was present in the detection range) or OFF (no vehicle present in the detection range. This occupancy state is internally stored in the sensor node.
- step S5 it is determined whether a signal value v obtained from the A/D converter, and indicative for an occupied status of the detection area is below a first predetermined value T L . If this is not the case program flow continues with step S2. If however the value is lower than said first predetermined value then program flow continues with step S6. In step S6 it is verified whether the signal value v remains below the first predetermined value T L for a first predetermined time period. During step S6 the retrieval of input from the A/D convertor is continued. If the signal value v returns to a value higher then said predetermined value T L before the end of said predetermined time-period then processing flow continues with step S2. Otherwise the value for the occupancy state is internally saved as unoccupied in step S7, and a message signaling this is transmitted in step S8.
- step S9 it is determined whether the signal value v obtained from the A/D converter, and indicative for an occupied status of the detection area is above a second predetermined value T H .
- the second predetermined value T H is higher than the first predetermined value T L . If this is not the case program flow continues with step S2. If however the value is higher than said second predetermined value T H then program flow continues with step S10.
- step S10 it is verified whether the signal value v remains above the second predetermined value T H for a second predetermined time period, which may be equal to the first predetermined time period. During step S10 the retrieval of input from the A/D convertor is continued.
- step S2 If the signal value v returns to a value lower then said predetermined value T H before the end of said predetermined time-period then processing flow continues with step S2. Otherwise the value for the occupancy state is internally saved as occupied in step S11, and a message signaling this is transmitted in step S12.
- FIG 8 illustrates the signal flow in a message interpreter MI.
- a radio receiver 20 receives the binary "vehicle present" signals D (optionally with timestamp) from the sensor nodes 10 via the radio link and runs a model based state estimator algorithm to calculate the motion states of the vehicles individually (i.e. each vehicle is represented in the message interpreter).
- the sensor density may be chosen dependent on the required accuracy of the estimation. If a very accurate vehicle tracking is desired multiple sensors per vehicle area may be present.
- the message interpreter MI has a vehicle database facility 32, 34 that comprises state information of vehicles present at the vehicle infrastructure.
- the message interpreter MI further has a sensor map 45describing the spatial location of the sensor nodes 10.
- the sensor nodes may transmit their location, or their position could even be derived by a localization method for wireless sensor networks.
- the message interpreter MI further has an association facility 40 for associating the messages D provided by the sensor nodes 10 with the state information present in the vehicle data base facility 32, 34.
- the association facility 40 may associate the messages received with state information for example with one of the methods Gating, Nearest Neighbor (NN), (Joint) Probabilistic Data Association ((J)DPA), Multiple Hypothesis Tracker (MHT) and the MCMCDA.
- the message interpreter further has a state updating facility 50 for updating the state information on the basis of the messages D associated therewith by the association facility 40. Once the messages D are associated with a particular vehicle the state of that vehicle in a local vehicle data base is updated by the state updating facility 50.
- the association facility 40 and the state updating facility 50 together form a database updating facility DBU.
- a global map builder 65 may exchange this updated information with global map builders of neighboring message interpreters via network interface 60 (wired or wireless) and to receive close to border detections.
- network interface 60 wireless or wireless
- Other uses are also possible to exchange the motion state of crossing vehicles (e.g. to calculate system level features like vehicle density and average velocity, but these are independent from the motion state estimation).
- a message interpreter MI shown in Figure 9 , consists of a radio receiver 20, coupled to antenna 22, a processing unit 24 (with memory 28) and a network interface 65, as well as a real-time clock 26.
- a real-time clock may be part of the sensor node, and the sensor node may embed a time-stamp indicative for time at which an event was detected in the message.
- a message interpretor will have a more reliable clock, as it can be more reliable synchronized with a reference clock.
- the network interface 65 couples the message interpreter MI via the communication channel 60 to other message interpreters.
- the microcontroller 24 of Figure 9 processes the received messages D.
- the memory 28 stores the local and global vehicle map and the sensor map as well as the software for carrying out the data association and state estimation tasks.
- separate memories may be present for storing each of these maps and for storing the software.
- dedicated hardware may be present to perform one or more of these tasks.
- the result of the processing i.e. the estimation of the motion states of all sensed vehicles
- the result of the processing is present in the memory of the message interpreters in a distributed way.
- Message interpreters may run additional (cooperative) algorithms to deduct higher level motion characteristics and/or estimate additional vehicle characteristics (e.g. geometry).
- the vehicle tracking system may comprise only a single message interpreter MI.
- MI message interpreter
- the global map builder is superfluous, and local vehicle map is identical to the global vehicle map.
- each message interpreter MI for a respective module comprises hardware as described with reference to Figure 8 and 9 .
- Figure 10 schematically shows a part of a vehicle infrastructure having sections R j-1 , R j , R j+1 .
- a vehicle moves in a direction indicated by arrow X from R j-1 , via R j , to R j+1 .
- Figure 11 shows an overview of a method for detecting the vehicle performed by the message interpreter for section R j , using the messages obtained from the sensor nodes.
- step S20 the method waits for a message D from a sensor node.
- program flow continues with step S21, where the time t associated with the message is registered.
- the registered time t associated with the message may be a time-stamp embedded in the message or a time read from an internal clock of the message interpreter.
- step S22 it is verified whether the detection is made by a sensor node in a location of section Rj that neighbors one of the neighboring sections R j-1 or R j+1 . If that is the case, then in step S23 the event is communicated via the communication network interface to the message interpreter for that neighboring section.
- step S24 it is determined which vehicle O in the vehicle data base facility is responsible for the detected event. An embodiment of a method used to carry out step S24 is described in more detail in Figure 12 . After the responsible vehicle O is identified in Step 25, i.e. an association is made with existing vehicle state information, it is determined in Step 26 whether it is present in the section Rj. If that is the case, control flow continues with Step S27, where the state of vehicle O is estimated.
- step S28 it is determined whether the state information implies that the vehicle O has a position in a neighboring section R j-1 or R j+1 . In that case the updated state information is transmitted in step S29 to the message interpreter for the neighboring section and control flow returns to step S20. Otherwise the control flow returns immediately to Step S20.
- the current state known for the vehicle with that index i is retrieved from the vehicle database facility.
- a probability is determined that the vehicle O caused the detection reported by the message D at time t.
- the vehicle index i is incremented in step S43 and if it is determined in step S44 that i is less than the number of vehicles, the steps S41 to S43 are repeated. Otherwise in step S45 it is determined which vehicle caused the detection reported by the message D at time t with the highest probability.
- the index of that vehicle is returned as the result if the method.
- step S60 the messages D 1 ,...,D n associated with vehicle O are selected.
- step S61 a probability density function is constructed on the basis of the associated messages D 1 ,...,D n .
- step S62 the current state S 0 and time to for vehicle O is retrieved from the vehicle database.
- step S63 it is determined whether the time for which the state S of the vehicle O has to be determined is greater than the time to associated with the current state S 0 .
- the state S (determined by the estimation method) is the state update of S0 to t, performed in step S65. If that is not the case, then the message D relates to a detection preceding the detection that resulted in the earlier estimation for state S0. In that case the state S0 is updated using the detection D by the state estimation method in step S64
- A1 Estimation and association for multiple target tracking based on spatially, distributed detections
- multiple target tracking [1-3] one aims to track all the objects/targets, which are moving in a certain area.
- Section 2 defines background knowledge such as the notation of (object) variables and functions that are used throughout this paper. After that the problem is formulated in section 3 together with existing methods. Section 5 describes the approach which is taken in the design. A more detailed description of the estimation and associated is presented in Section 5 and 6 respectively. Finally both methods are tested in a small application example presented in Section 6 and conclusions are drawn in section 7. But let's start with the background information.
- R defines the set of real numbers whereas the set R + defines the non-negative real numbers.
- the set Z defines the integer values and Z + defines the set of non-negative integer numbers.
- the variable 0 is used either as null, the null-vector or the null-matrix. Its size will become clear from the context.
- Vector x ( t ) ⁇ R n is defined as a vector depending on time t and is sampled using some sampling method.
- the time t at sampling instant k ⁇ Z + is defined as t k ⁇ R.
- the matrix A ( t 2 -t 1 ) ⁇ R a ⁇ b depends on the difference between two time instants t 2 > t 1 and is shortly denotes as A t 2 - t 1
- each object also has a certain shape or geometry which covers a certain set of positions in R xy , i.e. the grey area of Figure 14 .
- To define the vectors ⁇ i we equidistant sample the rectangular box defined by C 0 using a grid with a distance r .
- Each ⁇ i is a grid point within the set S as graphically depicted in Figure 15 .
- a total of E objects are observed within the set R xy .
- T i represents the i th object's rotation-matrix dependent on ⁇ i .
- the objects are observed in R xy by a camera or a network of sensors. For that M 'detection' points are marked within R xy and collected in the set D ⁇ R xy .
- the position of a detection point is denoted as d ⁇ D.
- Figure 16 shows an example of object i which is detected by multiple detection points. The covariance ⁇ of each detection point is also indicated.
- the sampling method of the observation vectors z 0:k is a form of event sampling [4, 5, 7]. For a new observation vector is sampled whenever an event, i.e. object detection, takes place. With these event samples all N objects are to be tracked. To accomplish that three methods are needed. The first one is the association of the new observation-vector z k to an object i and therefore denote it with z k i . Suppose that all associated observation-vectors z n i are collected in the set Z k i ⁇ z 0 : k . Then the second method is to estimate m k i from the observation-set Z k i . This is used in the third method, which is a state-estimator.
- Z k i is defined as the set with all observation-vectors from z 0: k that were associated with object i .
- Z k i is defined as the set with all observation-vectors from z 0: k that were associated with object i .
- the set Z k i ⁇ z 0 : k is defined as the set of all observation-vectors z n which were associated with object i , from which their detection point is still covered by the object. We will first show how this is done. At time step k we have the observation-set Z k ⁇ 1 i and the observation z k was associated to object i , i.e. z k i . Now if the object's edge is detected at d k for the first time, then z k i is added to the set Z k ⁇ 1 i .
- Estimation of the measurement-vector m k i given the observation set Z k i results in calculating p m k i
- the set Z consists of the observation vectors z n , for all n ⁇ N ⁇ [0, k ], that were associated to the same object.
- the main aspect of equation (17) is to determine p ( o
- O n ( ⁇ ) ⁇ R xy we define the set O n ( ⁇ ) ⁇ R xy to be equal to all possible object positions o, given that the object is detected at position d n ⁇ z n ( ⁇ Z ) and that the object's rotation is equal to ⁇ .
- the determination of O n ( ⁇ ) ⁇ R xy is presented in the n the next section.
- Z, ⁇ ) and ⁇ l are related to the set O N ( ⁇ ) due to the fact that it O N ( theta ) defines the set of possible object positions o for a given ⁇ .
- O N ( theta ) defines the set of possible object positions o for a given ⁇ .
- Z , ⁇ ) f o
- z n , ⁇ : ⁇ 0 if o ⁇ O N ⁇ , 1 if o ⁇ O n ⁇ , g o
- z n , ⁇ ⁇ 0 if o ⁇ O N ⁇ , 1 if o ⁇ O N ⁇ ,
- Z ) is calculated according to (6).
- the rest of this section is divided into two parts. The first part derives the probability function based on a single detection, i.e. f ( o
- Figure 20 (right) graphically depicts the determination of ⁇ n from the set ⁇ for a given ⁇ and detection point d n .
- z n , ⁇ ) is approximated by placing a Gaussian function at each sampled position ô i ⁇ ⁇ n with a certain covariance dependent on the grid-size r : f o
- the aim of this section is to calculate the function g ( o
- Equation (22) If N contains m elements, then calculating equation (22) would result in K m products of m Gaussian functions and sum them afterwards. This would take too much processing power if m is large. That is why equation (22) is calculated differently.
- each detection point d n defines a rectangular set denoted with C n ( ⁇ ) dependent on rotation ⁇ .
- the intersection of all these rectangular sets is defined with the set C N ( ⁇ ) .
- the first set, O n ( ⁇ ),shown in Figure 17 defines all possible object positions o based on a single detection at d n .
- the second set, i.e. O N ( ⁇ ),shown in Figure 18 defines all possible object positions o based on all detections at d n , ⁇ n ⁇ N. Notice that as a result O n ( ⁇ ) ⁇ C n ( ⁇ ) and O N ( ⁇ ) ⁇ C N ( ⁇ ). Meaning that only within the set C N ( ⁇ ) all the functions f ( o
- Equation (25) is reduced to: g o
- Z , ⁇ ⁇ ⁇ n N 2 ⁇ ⁇ 2 ⁇ i ⁇ I n G o o ⁇ i n R , with N ⁇ N : C N ⁇ ⁇ O n ⁇ , ⁇ n ⁇ N ⁇ N .
- the calculation of (26) is done by applying the following two propositions.
- the first one i.e. Proposition 2
- the second one i.e. Proposition 3, proofs that a product of Gaussians results in a single Gaussian.
- Equation (29) is approximated as a single Gaussian function: g o
- Equation (30) is substituted into equation (16) together with f ( o
- Z ) also gives us the probability that a new observation vector is generated by an certain object i . This is discussed in the next section.
- the total probability that a new observation vector z k is generated by object i is equal to the total probability of the measurement-vector m k i given the observation set Z k ⁇ 1 i z k .
- Z k ⁇ 1 i , z k which is equal to equation (41).
- the definition of a PDF is that its total probability, i.e. its integral from - ⁇ to ⁇ , is equal to 1.
- Z k ⁇ 1 i , z k of equation (31) has a total probability of 1, it is divided by its true probability Pr m k i
- ⁇ i and K i are equal to ⁇ and K respectively, which define the approximation of the function f m k i
- the probability of (3) one can design a method which associates an observation-vector due to a new detection, to its most probable object i .
- the estimation method requires a certain amount of processing power, one can reduce this by reducing the number of samples in the set A. Meaning that association and estimation can be done with different sizes of A.
- the objects have a rectangular shape, then with some tricks one can reduce the amount of processing power to a level at which both association as well as estimation can run real-time.
- the simulation case is made such that it contains two interesting situation.
- the objects are tracked using two different association methods.
- the first one is a combination of Gating and detection association of 6.
- the second one is a combination of Gating and Nearest Neighbor.
- the result of the detection associator (DA) for both crossings is shown in Figure 22 while the result of the Nearest Neighbor (NN) associator is shown in Figure 23 .
- the real object is plotted in a thick, solid line while its estimated one is plotted in a thin, solid line.
- the associated detections of each object are given with a symbol which is different for each object; ' ⁇ ' if associated with vehicle 1, ' ⁇ ' if associated with vehicle 2, ' ⁇ ' if associated with vehicle 3 and '*' if associated with vehicle 4.
- Figure 22 shows with the DA all detections were correctly associated to the one object, while If NN is used as an association method,we see that a lot of incorrect associated detections. Therefore we can concluded that using the detection association of 7 results in less estimation-error compared to NN.
- This paper presents a method for estimating the position- and rotation-vector of objects from spatially, distributed detections of that object. Each detection is generated at the event that the edge of an object crosses a detection point. From the estimation method a detection associator is also designed. This association method calculates the probability that a new detection was generated by an object i.
- An example of a parking lot shows that the detection association method has no incorrect associated detections in the case that two vehicles cross each other both in parallel as well as orthogonal. If the association method of Nearest Neighbor was used, a large amount of incorrect associated detections were noticed, resulting in a higher state-estimation error.
- the data-assimilation can be further improved with two adjustments.
- the first one is replacing the set S with S E only at the time-instants that the observation vector is received.
- the second improvement is to take the detection points that have not detected anything also in account.
- x p x dx 34 a ⁇ ⁇ ⁇ ⁇ G m , ⁇ x , M G x u U dx . 34 b
- T o reduce the amount of data transfer in networked control systems and wireless sensor networks, measurements are usually taken only when an event occurs, rather that at each synchronous sampling instant. However, this complicates estimation and control problems considerably.
- the goal of this paper is to develop a state estimation algorithm that can successfully cope with event based measurements.
- we develop a state estimator with a hybrid update i.e. when an event occurs the estimated state is updated using measurements; otherwise the update is based on the knowledge that the monitored variable is within a bounded set used to define the event.
- a sum of Gaussians approach is employed to obtain a computationally tractable algorithm.
- R defines the set of real numbers whereas the set R + defines the non-negative real numbers.
- the set Z defines the integer numbers and Z + defines the set of non-negative integer numbers.
- the notation 0 is used to denote either the null-vector or the null-matrix. Its size will become clear from the context.
- a vector x ( t ) ⁇ R n is defined to depend on time t ⁇ R and is sampled using some sampling method. Two different sampling methods are discussed. The first one is time sampling in which samples are generated whenever time t equals some predefined value. This is either synchronous in time or asynchronous. In the synchronous case the time between two samples is constant and defined as t s ⁇ R + .
- the second sampling method is event sampling, in which samples are taken when an event occurred.
- the i th and maximum eigenvalue of a square matrix A are denoted as ⁇ i ( A ) and ⁇ max ( A ) respectively.
- a ⁇ R n ⁇ n and B ⁇ R n ⁇ n are positive definite, denoted with A ⁇ 0 and B ⁇ 0, then A ⁇ B denotes A - B ⁇ 0.
- a ⁇ 0 denotes A is positive semi-definite.
- PDF probability density function
- the exact description of the set H k e (z k e -1 , t ) depends on the actual sampling method. As an example H k e (z k e - 1 , t ) is derived for the method "Send-on-Delta", with y ( t ) ⁇ R.
- H k e ( z k e - 1 , t ) should contain the set of all possible values that y ( t ) can take in between the event instants k e - 1 and k e . Meaning that if t k e -1 ⁇ t ⁇ t k e , then y ( t ) ⁇ H k e ( z k e -1 , t ) .
- a sufficient condition is that y k e -1 ⁇ H k e ( z k e -1 ,t ) , which for "Send-on-Delta" results in y ( t ) ⁇ [ y k e -1 - ⁇ , y e -1 + ⁇ ] for all t k e -1 ⁇ t ⁇ t k e .
- the state vector x(t) of this system is to be estimated from the observation vectors z 0 e : k e .
- our goal is to construct an event-based state-estimator (EBSE) that provides an estimate of x(t) not only at the event instants t k e but also at the sampling instants t k a . Therefore, we define a new set of sampling instants t n as the combination of sampling instants due to event sampling, i.e.
- the estimator calculates the PDF of the state-vector x n given all the observations until t n . This results in a hybrid state-estimator, for at time t n an event can either occur or not, which further implies that measurement data is received or not, respectively. In both cases the estimated state must be updated (not predicted) with all information until t n .
- the PDFs of (9) can be described as the Gaussian G ( x n, x n
- n i.e. ⁇ i P n
- the problem of interest in this paper is to construct a state-estimator suitable for the general event sampling method introduced in Section 3 and which is computationally tractable. Furthermore, it is desirable to guarantee that P n
- Existing state estimators can be divided into two categories.
- the first one contains estimators based on time sampling: the (a)synchronous Kalman filter [12, 13] (linear process, Gaussian PDF), the Particle filter [14] and the Gaussian sum filter [4, 5] (nonlinear process, non-Gaussian PDF).
- These estimators cannot be directly employed in event based sampling as if no new observation vector z k e is received, then t n -t k e ⁇ ⁇ and ⁇ i (P n
- the second category contains estimators based on event sampling. In fact, to the best of our knowledge, only the method proposed in [15] fits this category.
- Equation (25) is explicitly solved by applying Proposition 1: p x n
- n ⁇ 1 + C T R n i ⁇ 1 y n i , P n i : P n
- n ⁇ 1 ⁇ 1 + C T R n i ⁇ 1 C ⁇ 1 and ⁇ n i : G y n i , Cx n
- the third step is to approximate (27) as a single Gaussian to retrieve a computationally tractable algorithm. For if both p ( x n -1
- y 0: n ⁇ Y 0: n ) of (27) is approximated as a single Gaussian with an equal expectation and covariance matrix, i.e.: p x n
- the first two estimators are the EBSE and the asynchronous Kalman filter (AKF) of [13].
- the AKF estimates the states only at the event instants t k e .
- the states at t k a are calculated by applying the prediction-step of (14b).
- the third estimator is based on the quantized Kalman filter (QKF) introduced in [21] that uses synchronous time sampling of y k a .
- the QKF can deal with quantized data, which also results in less data transfer, and therefore can be considered as an alternative to EBSE.
- y k a is the quantized version of y k a with quantization level 0.1, which corresponds to the "Send-on-Delta" method. Hence, a comparison can be made.
- ⁇ ⁇ R + x i ⁇ x i
- i ⁇ 1 , which is a measure of the change in the estimation-error after the measurement update with either z k e or y k a was done. Notice that if ⁇ ⁇ 1 the estimation error decreased after an update, if ⁇ > 1 the error increased and if ⁇ 1 the error remained the same.
- the last aspect on which the three estimators are compared is the total amount of processing time which was needed to estimate all state-vectors.
- both x k e and x k a were estimated and it took 0.094 seconds.
- the EBSE calculates P n
- n 2 as (39) in which V is replaced with R : V + V . Notice that for these estimators we have that ⁇ n ⁇ t s and R n ° R , for all n . Let the EBSE and the KF start with the same initial covariance matrix P 0 . The first step of induction is to prove that P 1
- 1 1 A ⁇ 1 P 0 A ⁇ 1 T + B ⁇ 1 QB ⁇ 1 T ⁇ 1 + C T R 1 ⁇ 1 C ⁇ 1
- 1 2 A t s P 0 A t s T + B t s QB t s T ⁇ 1 + C T R ⁇ 1 C ⁇ 1 .
- V 1 : A ⁇ 1 P 0 A ⁇ 1 T + B ⁇ 1 QB ⁇ 1 T
- V 2 : A t s P 0 A t s T + B t s QB t s T
- 1 1 and U 2 : P 1
- the second and last step of induction is to show that if P n ⁇ 1
- V 1 : A ⁇ n P n ⁇ 1
- V 2 : A t s P n ⁇ 1
- the second condition of Lemma 6.4, i.e. V 1 ° V 2 also holds by applying Lemma 6.4, i.e. A ⁇ n P n ⁇ 1
- n 1 and U 2 : P n
Landscapes
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Traffic Control Systems (AREA)
- Radar Systems Or Details Thereof (AREA)
Description
- Vehicle tracking system, vehicle infrastructure provided with vehicle tracking system and method for tracking vehicles.
- The present invention relates to a vehicle tracking system.
- The present invention relates to a vehicle infrastructure provided with a vehicle tracking system.
- The present invention further relates to a method for tracking vehicles.
- In "Measuring Traffic", Statistical Science, 2007, Vol. 22, No. 4, pp. 581-597, Peter J. Bickel et al. describe a traffic performance measurement system, PeMS, that functions as a statewide repository for traffic data gathered by thousands of automatic sensors. It has integrated data collection, processing and communications infrastructure with data storage and analytical tools. This system provides for global information relating to the average traffic speed and the traffic density.
- There is a need for a system that provides information relating to the state of individual traffic participants. This information can be used in advanced cruise controllers for traffic management or for access control for example.
-
US5801943 describes a wide area surveillance system for application to large road networks. The system employs smart sensors to identify plural individual vehicles in the network. These vehicles are tracked on an individual basis, and the system derives the behavior of the vehicle. Furthermore, the system derives traffic behavior on a local basis, across roadway links, and in sections of the network. Processing in the system is divided into multiple processing layers, with geographical separation of tasks. - The vehicle tracking system according to the present invention comprises
- a plurality of sensor nodes that each provide a message indicative for an occupancy status of a detection area of a vehicle infrastructure monitored by said sensor node,
- a message interpretator including a vehicle database facility with state information of vehicles present at the vehicle infrastructure, and a database updating facility for updating the database facility on the basis of messages provided by the sensor nodes, characterized in that said sensor nodes are arranged in the vehicle infrastructure at a density of at least 0.2 per square meter.
- Contrary to the system known from
US5801943 , in the vehicle tracking system according to the present invention vehicles can be tracked with relatively simple and cheap means. Smart sensors are not necessary. It is sufficient that the sensor nodes sense an occupancy state, i.e. whether a detection area associated with the sensor node is occupied by a vehicle or not and that they merely provide a message that indicates whether the occupancy state is changed. The relatively cheap and simple construction of the sensor nodes contributes to an economically feasible application in vehicle tracking systems for large vehicle infrastructures. Optionally, the message may additionally include the value of the occupancy state after the change was detected. The plurality of sensor nodes arranged in the vehicle infrastructure having at least the above-mentioned density provide a course image of the vehicles present at the vehicle infrastructure. However, as compared to an image provided by cameras, the image provided by the plurality of sensor nodes is always captured from the same perspective. This facilitates further processing. - In an embodiment the database update facility comprises
- an association facility for associating the messages provided by the sensor nodes with the state information present in the vehicle data base facility,
- a state updating facility for updating the state information on the basis of the messages associated therewith.
- The association facility selects for which state information the received messages are relevant, and provides the selected messages to the state updating facility. In this way the state updating facility can operate more efficiently, than in case no selection takes place.
- The present invention is in particular suitable for tracking vehicles. For tracking vehicles, suitable sensor elements to be used in the sensor nodes are for example magneto restrictive sensors. These sensors determine whether their associated detection area is occupied by detection of a perturbation of the earth magnetic field. Alternatively, magnetic loop sensors, may be used, which detect a change of inductance caused by the presence of ferromagnetic material.
- Preferably each sensor node is provided with a wireless transmission facility that transmits the preprocessed data, e.g. the occupance status or an indication of a change thereof to a data to a receiver facility coupled to the message interpreter. The absence of wiring towards the message interpreter makes the installation easier and cost effective. Furthermore it is attractive if the sensor nodes provide their message at an event basis, e.g. if a perturbation of the earth magnetic field exceeds a threshold value. This reduces communication load of the message interpreter and minimizes power consumption of the sensor nodes.
- In an embodiment the density of the sensor nodes are at least 0.6 per square meter. Due to the relatively high density of the sensor nodes in the traffic infrastructure in said embodiment an individual failure of a sensor or of an individual sensor does not have serious consequences for the estimation of the states of the traffic participants. Accordingly, any return transmission from the message interpreter to the sensor nodes to actively verify the occupancy state is superfluous, which is also favorable for a low power consumption of the sensor nodes.
- The vehicle tracking system may in addition to the plurality of sensor nodes arranged in the traffic infrastructure comprise one or more cameras. A camera may be used for example if a perturbation of the earth magnetic field can not be measured. This is the case for example if (parts of) the infra structure comprises metal components e.g. a bridge.
It is not necessary that the detection areas of the sensor elements are complementary. The detection areas may overlap, or spaces may exist between the detection areas, but it is required that the detection area of the sensor be smaller than the vehicles to be tracked. Ideally the sensor elements are point detectors. - The sensor nodes can be either randomly distributed over the vehicle infrastructure or placed in a pattern optimized for the vehicle tracking problem in hand.
- In an embodiment the vehicle tracking system comprises a plurality of system modules, each module comprising a respective subset of the plurality of sensor nodes for monitoring a respective section of the vehicle infrastructure and a respective message interpreter, the vehicle tracking system has a communication facility for enabling system modules of mutually neighboring sections to exchange state and detection information. In this way the vehicle tracking system can be easily expanded if required. A new system module need only to communicate with the system modules arranged for neighboring sections. Neighboring sections may be arranged in one dimensional scheme, e.g. in case of a narrow road. For example if a certain road is already provided with an vehicle tracking system, it is sufficient to provide for a communication facility between the system module for the last section of said vehicle tracking system and the new system module for the appended section. In case of infrastructures for unconstrained vehicle movements the new module may communicate with other modules neighboring in various directions. As the system modules merely need to exchange state information and vehicle-detection information (i.e. the unprocessed sensor signals) in a limited subarea of the respective sections, the amount of communication between the system modules is modest resulting in a scalable vehicle tracking system.
- The association facility associates the messages provided by the sensor nodes or neighboring system modules with the state information present in the vehicle data base facility. In other words the association facility determines the probability that the detections are caused by a particular vehicle for which state information is present in the vehicle data base facility. If the messages cannot be associated with state information of an already identified vehicle here or in the neighboring system module, a new entry may be added to the database. Alternatively, the entry for the new vehicle may be added by a separate procedure. For example the vehicle infrastructure may have an access with a vehicle identification facility that provides for an identification of every vehicle that enters the infrastructure.
- The individual sensor nodes do not need to provide other information than an occupancy status of their associated detection area. However, optionally the sensor node may associate its own signal with a color, shape, or other signature of the tracked vehicles to facilitate or obviate association by the message interpreter. An association facility for associating the detection signals obtained on asynchronous basis with state information of a particular vehicle may be based on one of the following methods.
- Gating,
- Nearest Neighbor (NN),
- (Joint) Probabilistic Data Association ((J)DPA),
- Multiple Hypothesis Tracker (MHT), and
- MCMCDA.
- Gating comprises forming a gate around the predicted measurement of a vehicle. The size and shape of the gate are chosen in such a way that unlikely messages are precluded to be associated with this particular vehicle-track. The method determines a statistical, quadratic
distance - Various methods can be used for finding the Threshold G. However, this data association method is not suitable for associating event based messages. Moreover problems arise when two gates overlap.
- The Nearest Neighbor method also uses a gate, but it can handle overlapping gates. The sum of all possible combinations to associate a certain measurement to a certain track is analyzed. The chosen combination associates the most measurements to a track for a minimum sum of distances.
- A (Joint) Probabilistic Data Association method is described in Multitarget-Multisensor Tracking: Principles and Techniques. YBS, 1995, by Y. Bar-Shalom and R. Li, for example. The (Joint) Probabilistic Data Association methods (J)PDA is unsuitable for event-based sampling because it assumes that one target can give rise to at most one measurement and one measurement is a result of at most one vehicle. This cannot be assumed with event based sampling. An extension to the JPDA can be found in O. Songhwai, S. Sastry, and L. Schenato, "A Hierarchical Multiple-Target Tracking Algorithm for Sensor Networks," in Proc. of the 2005 IEEE International Conference on Robotics and Automation, Barcelona, Spain, 2005. Where the original JDPA assumes that a probability density function (PDF) of state-vector xk+1|jk is a Gaussian function, in the extended JPDA the PDF can be a non-Gaussian function.
- A Multiple Hypothesis Tracker (MHT) allows that the state-vector of a single vehicle can has multiple tracks. This method resembles to the Particle filter as described in B. Ristic, S. arulampalam, and N. Gordon, "Beyond the Kalman filter: Particle filter for tracking applications", 2002. Therein, each state is estimated by simulating N states with each a different probability. It is a drawback of this method that it requires a high computational power.
- A further data association method is the Markov chain Monte Carlo data association (MCMCDA). All observations are used to classify and cluster them. To that end the whole set of observations is divided into a number of partitions represented by the set w. This is done nmc times resulting in nmc sets of w, i.e. possible partitions. The set of w with the highest probability, given the number of vehicles in the previous sample instant, is chosen and the state-vectors of the tracks a are updated according the partitioned observation. The computational time can be decreased by not using the total history of observations, but by using a moving horizon. A downside of this method is that each observation can belong to at most one vehicle and, making this method unsuitable for event-based state-estimation.
- Although these data association methods can be applied to associate messages obtained by (a)synchronous sampling, they are not suitable for association messages obtained by event-based sampling. For then the main issue with present association methods is that they assume either that one target results in at most one observation or that one observation comes from at most one vehicle. According to a preferred embodiment of the present invention messages received from the sensor nodes are first associated with a particular vehicle. In other words it is determined which vehicle most probably caused the observation by the sensor resulting in the message. Subsequently the state of said vehicle is (re-)estimated using that message. The estimation is based on the latest observation and the current state of the said vehicle. The state of the vehicle that is most probable in view of all available observations is calculated.
- More in particular the step of associating may comprise
- initializing (S40) a vehicle index (i),
- retrieving (S41) the current state known for the vehicle with that index from a vehicle database facility,
- determining (S42) a probability that the vehicle with that index caused the detection reported by the message D,
- incrementing (S43) the vehicle index,
- determining (S44) whether the vehicle index is less than the number of vehicles,
- if the outcome of the determination is positive repeating steps S41 to S43 with the incremented vehicle index, and
- if the outcome of the determination is negative, determining (S45) which vehicle caused the detection reported by the message D with the highest probability.
- returning (S46) the index of the vehicle identified in step S45.
- If no message is received, e.g. because the transmission of the message was disturbed by external causes, or because the sensor node is defect, there can accordingly be no association. However, due to the relatively high density of sensor nodes, it is not necessary to verify whether the absence of a message is due to a failure of the message creation and transmission or due to the absence of a vehicle. I.e. the message interpreter does not need to send verification messages to the sensor nodes to verify correct operation of the transmission. Accordingly a one-way message traffic from the sensor nodes to the message interpreters is sufficient for a correct operation of the method.
- The state of a vehicle can also be estimated at a point in
time later than the last message, but before a new message has arrived. In that case the error covariance matrix is bounded, as it is known that the state change of the vehicle must be within the detection boundaries of the sensor node. - These and other aspects are described in more detail with reference to the drawing. Therein:
-
Figure 1 shows a first view of an embodiment of a vehicle infrastructure provided with a vehicle tracking system according to the invention, -
Figure 2 shows a second view of an embodiment of a vehicle infrastructure provided with a vehicle tracking system according to the invention, -
Figure 3 shows another view of an embodiment of a vehicle tracking system according to the invention, -
Figure 4 schematically shows a part of the vehicle infrastructure that is provided with a plurality of sensor nodes, -
Figure 5 schematically shows a signal flow in a sensor node, -
Figure 6 schematically shows a possible hardware implementation of a sensor node, -
Figure 7 shows a possible method carried out by a sensor node, -
Figure 8 shows a signal flow in a message interpretor, -
Figure 9 shows a possible hardware implementation of a message interpretor, -
Figure 10 shows neighboring infrastructure regions, with specially handled subregions marked, -
Figure 11 shows an overview of a method carried out by the message interpreter, -
Figure 12 shows a first detail of the method ofFigure 11 , -
Figure 13 shows a second detail of the method ofFigure 11 , -
Figure 14 shows an example of a vehicle to be detected at a reference position and orientation and at a different position and orientation, -
Figure 15 shows a definition of a set S and the equidistant sampled set A, -
Figure 16 shows detection of a vehicle at multiple detection points, -
Figure 17 shows a definition of the set On of possible positions oi k for a single detection point, -
Figure 18 shows a definition of the set ON of possible positions oi k for multiple detection points, -
Figure 19 shows a derivation of ON(q) given 2 detections and 2 different samples of q, -
Figure 20 shows (Left) determination of -A, (right) the vehicle's possible position set ^On given dn and q, -
Figure 21 shows (left) the mean of all Gaussians from f(o |z1, θ) and f(o|z2, θ); (right) The selection of means of the Gaussians from f(o |z1, θ) and f(o |z2, θ), of which their mean ôni is close or in CN(θ), -
Figure 22 shows an association result with event-based data-association, -
Figure 23 shows an association result with Nearest Neighbor data-association, -
Figure 24 shows time sampling of a signal y(t), -
Figure 25 shows event sampling of a signal y(t), -
Figure 26 shows event sampling: Send-on-Delta, -
Figure 27 shows the Gaussian function, -
Figure 28 shows a top view of the Gaussian function, -
Figure 29 shows an approximation of Λ Hke (yn ) as a sum of Gaussian functions, -
Figure 30 shows position, speed and acceleration of a simulated vehicle, -
Figure 31 shows a position estimation error for various methods, -
Figure 32 shows a speed estimation speed for various methods, -
Figure 33 shows a factor of increase in estimation error after zke , ory ka . - In the following detailed description numerous specific details are set forth in order to provide a thorough understanding of the present invention. However, it will be understood by one skilled in the art that the present invention may be practiced without these specific details. In other instances, well known methods, procedures, and components have not been described in detail so as not to obscure aspects of the present invention.
- The invention is described more fully hereinafter with reference to the accompanying drawings, in which embodiments of the invention are shown. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art. In the drawings, the size and relative sizes of layers and regions may be exaggerated for clarity.
- It will be understood that when an element or layer is referred to as being "on", "connected to" or "coupled to" another element or layer, it can be directly on, connected or coupled to the other element or layer or intervening elements or layers may be present. In contrast, when an element is referred to as being "directly on," "directly connected to" or "directly coupled to" another element or layer, there are no intervening elements or layers present. Like numbers refer to like elements throughout. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
- It will be understood that, although the terms first, second, third etc. may be used herein to describe various elements, components, and/or sections, these elements, components, and/or sections should not be limited by these terms. These terms are only used to distinguish one element, component or section from another element, component, and/or section. Thus, a first element, component, and/or section discussed below could be termed a second element, component, and/or section without departing from the teachings of the present invention.
- Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein. In addition, the materials, methods, and examples are illustrative only and not intended to be limiting.
-
Figure 1 and 2 show a first and a second view of an embodiment of avehicle infrastructure 80 provided with a vehicle tracking system. The vehicle infrastructure is intended to allow stationary and/or movingvehicles 70 thereon, e.g. a road or a parking place. The vehicle infrastructure may be part of a public or private space, e.g. a recreational park. - The vehicle tracking system comprises a plurality of
sensor nodes 10 that each provide a message indicative for an occupancy status of a detection area of the vehicle infrastructure monitored by saidsensor node 10. As shown therein the sensor nodes are randomly distributed over the vehicle infrastructure. - The vehicle tracking system comprises a message interpretator MI, each comprising a vehicle database facility, an association facility and a state updating facility. Each message interpretator is responsible for handling messages D from a
respective section vehicle infrastructure 80. -
Figure 3 is another schematic view of the vehicle tracking system.Figure 3 shows howsensor nodes 10 transmit (detection) messages to a message interpreter MI in their neighborhood. The message interpreters MI may also communicate to each other via acommunication channel 60 to indicate that a vehicle crosses a boundary between their respective sections and to exchange a status of such a vehicle. As shown inFigure 3 , the vehicle tracking system comprises a plurality of system modules MD1, MD2, MD3. Although three modules are shown in this example, any number of
system modules is possible, dependent on the application. For example for an isolated vehicle infra structure, e.g. an intersection of roads a single module may be applicable, while on a long road thousands of modules may be present. Eachmodule MD 1, MD2, MD3 comprises a respective subset of the plurality ofsensor nodes 10 for monitoring a respective section of the vehicle infrastructure and a respective message interpreter MI. The vehicle tracking system further has acommunication facility 60 for enabling system modules MD1, MD2, MD3 of mutually neighboring sections to exchange state information. In the embodiment shown, messages from the sensor nodes are directly transmitted to a message interpretor. Alternatively, the sensor nodes may form a network that routes the messages to the message interpreters. In that case the transmitters may have a short transmission range. -
Figure 4 schematically shows a part of the vehicle infrastructure that is provided with a plurality of sensor nodes j having position cj. The sensor nodes have a detection area with radius R. A vehicle i is present at the infrastructure having a position (vi x, vi y). In this case if the vehicle substantially covers the detection area the sensor node indicates that the detection area is occupied as indicated in gray. Otherwise the sensor node indicates that the detection area is not occupied (white). In practice the fraction of the detection area that should be covered before an occupied status is detected may deviate from the above-mentioned 50% depending on the type of vehicle. -
Figure 5 schematically illustrates the signal flow for thesensor node 10, havingsensor element 12, a processing unit 14 (with memory), and aradio link 16. - The
sensor element 12 is capable of sensing the proximity of the vehicles to be tracked. Theprocessing unit 14 determines if a vehicle is present or absent on the basis of the signals from thesensor element 12. If an occupancy status of the detection area of the sensor changes, theprocessing unit 14 initiates a transmission of a message D indicating the new occupancy status. In an alternative embodiment the message may include a time stamp indicative of the time t at which the new occupancy status occurred. Alternatively the sensor nodes may transmit occupancy status information on a periodical basis for example. However, an event-based transmission enables a lower power use. The message D sent should reach at least one message interpreter MI. In a concrete implementation of thesensor node 10 for tracking vehicles on a road thesensor element 12 is a magnetoresistive component, which measures the disturbance on the earth magnetic field induced by the vehicles. Alternatively, a magnetic rod or loop antenna may be used to detect the occupancy by a vehicle. -
Figure 6 shows a possible implementation of the hardware involved for thesensor node 10 ofFigure 5 . Thesensor element 12 is coupled via an A/D converter 13 to amicrocontroller 14 that has access to amemory 15, and that further controls aradio transmitter 16 coupled to anantenna 17. -
Figure 7 schematically shows a method performed by a sensor node to generate a message indicative for occupancy status of a detection area of the sensor node. - Starting (Step S1: initialization) from an off-state of the sensor node, input from the A/D converter is received (Step S2). In a next step S3, offset is removed from the sensed value.
- In step S4 it is determined whether the occupancy state of the detection area as reported by the last message transmitted by the sensor node was ON (vehicle was present in the detection range) or OFF (no vehicle present in the detection range. This occupancy state is internally stored in the sensor node.
- In the first case, program flow continues with step S5. In the second case processing flow continues with step S9. In step S5 it is determined whether a signal value v obtained from the A/D converter, and indicative for an occupied status of the detection area is below a first predetermined value TL. If this is not the case program flow continues with step S2. If however the value is lower than said first predetermined value then program flow continues with step S6. In step S6 it is verified whether the signal value v remains below the first predetermined value TL for a first predetermined time period. During step S6 the retrieval of input from the A/D convertor is continued. If the signal value v returns to a value higher then said predetermined value TL before the end of said predetermined time-period then processing flow continues with step S2. Otherwise the value for the occupancy state is internally saved as unoccupied in step S7, and a message signaling this is transmitted in step S8.
- In step S9 it is determined whether the signal value v obtained from the A/D converter, and indicative for an occupied status of the detection area is above a second predetermined value TH. The second predetermined value TH is higher than the first predetermined value TL. If this is not the case program flow continues with step S2. If however the value is higher than said second predetermined value TH then program flow continues with step S10. In step S10 it is verified whether the signal value v remains above the second predetermined value TH for a second predetermined time period, which may be equal to the first predetermined time period. During step S10 the retrieval of input from the A/D convertor is continued. If the signal value v returns to a value lower then said predetermined value TH before the end of said predetermined time-period then processing flow continues with step S2. Otherwise the value for the occupancy state is internally saved as occupied in step S11, and a message signaling this is transmitted in step S12.
-
Figure 8 illustrates the signal flow in a message interpreter MI. As shown in more detail inFigure 8 , aradio receiver 20 receives the binary "vehicle present" signals D (optionally with timestamp) from thesensor nodes 10 via the radio link and runs a model based state estimator algorithm to calculate the motion states of the vehicles individually (i.e. each vehicle is represented in the message interpreter). The sensor density may be chosen dependent on the required accuracy of the estimation. If a very accurate vehicle tracking is desired multiple sensors per vehicle area may be present. - The message interpreter MI has a
vehicle database facility - The message interpreter MI further has a sensor map 45describing the spatial location of the
sensor nodes 10. Alternatively, the sensor nodes may transmit their location, or their position could even be derived by a localization method for wireless sensor networks. - The message interpreter MI further has an
association facility 40 for associating the messages D provided by thesensor nodes 10 with the state information present in the vehicledata base facility association facility 40 may associate the messages received with state information for example with one of the methods Gating, Nearest Neighbor (NN), (Joint) Probabilistic Data Association ((J)DPA), Multiple Hypothesis Tracker (MHT) and the MCMCDA. - The message interpreter further has a
state updating facility 50 for updating the state information on the basis of the messages D associated therewith by theassociation facility 40. Once the messages D are associated with a particular vehicle the state of that vehicle in a local vehicle data base is updated by thestate updating facility 50. - The
association facility 40 and thestate updating facility 50 together form a database updating facility DBU. - In the embodiment shown a
global map builder 65 may exchange this updated information with global map builders of neighboring message interpreters via network interface 60 (wired or wireless) and to receive close to border detections. Other uses are also possible to exchange the motion state of crossing vehicles (e.g. to calculate system level features like vehicle density and average velocity, but these are independent from the motion state estimation). - A message interpreter MI, shown in
Figure 9 , consists of aradio receiver 20, coupled toantenna 22, a processing unit 24 (with memory 28) and anetwork interface 65, as well as a real-time clock 26. In other embodiments a real-time clock may be part of the sensor node, and the sensor node may embed a time-stamp indicative for time at which an event was detected in the message. However, generally a message interpretor will have a more reliable clock, as it can be more reliable synchronized with a reference clock. Thenetwork interface 65 couples the message interpreter MI via thecommunication channel 60 to other message interpreters. - In the embodiment shown the
microcontroller 24 ofFigure 9 processes the received messages D. Thememory 28 stores the local and global vehicle map and the sensor map as well as the software for carrying out the data association and state estimation tasks. In an alternative embodiment separate memories may be present for storing each of these maps and for storing the software. Likewise dedicated hardware may be present to perform one or more of these tasks. - There is no communication or any other interaction between the vehicles tracked. The result of the processing (i.e. the estimation of the motion states of all sensed vehicles) is present in the memory of the message interpreters in a distributed way.
Message interpreters may run additional (cooperative) algorithms to deduct higher level motion characteristics and/or estimate additional vehicle characteristics (e.g. geometry). - For applications in relative small area, e.g. a parking place, the vehicle tracking system may comprise only a single message interpreter MI. In that case the global map builder is superfluous, and local vehicle map is identical to the global vehicle map.
- In the embodiment shown in
Figure 3 , each message interpreter MI for a respective module comprises hardware as described with reference toFigure 8 and9 . - Operation of the message interpreter is further illustrated with respect to
Figures 10-13 -
Figure 10 schematically shows a part of a vehicle infrastructure having sections Rj-1, Rj, Rj+1. By way of example it is presumed that a vehicle moves in a direction indicated by arrow X from Rj-1, via Rj, to Rj+1. -
Figure 11 shows an overview of a method for detecting the vehicle performed by the message interpreter for section Rj, using the messages obtained from the sensor nodes. - In step S20 the method waits for a message D from a sensor node. At the moment that a message D is received, program flow continues with step S21, where the time t associated with the message is registered. The registered time t associated with the message may be a time-stamp embedded in the message or a time read from an internal clock of the message interpreter.
- In step S22, it is verified whether the detection is made by a sensor node in a location of section Rj that neighbors one of the neighboring sections Rj-1 or Rj+1. If that is the case, then in step S23 the event is communicated via the communication network interface to the message interpreter for that neighboring section. In step S24 it is determined which vehicle O in the vehicle data base facility is responsible for the detected event. An embodiment of a method used to carry out step S24 is described in more detail in
Figure 12 . After the responsible vehicle O is identified in Step 25, i.e. an association is made with existing vehicle state information, it is determined inStep 26 whether it is present in the section Rj. If that is the case, control flow continues with Step S27, where the state of vehicle O is estimated. Otherwise control flow returns to step S20. A procedure for estimating the state is described in more detail with reference toFigure 13 . In step S28 it is determined whether the state information implies that the vehicle O has a position in a neighboring section Rj-1 or Rj+1. In that
case the updated state information is transmitted in step S29 to the message interpreter for the neighboring section and control flow returns to step S20. Otherwise the control flow returns immediately to Step S20. - A method to associate a message D at time t, with a vehicle O is now described in more detail with reference to
Figure 12 . - In a first step S40, a vehicle index i is initialized (e.g. i=1). In a next step S41, the current state known for the vehicle with that index i is retrieved from the vehicle database facility. In the next step S42 a probability is determined that the vehicle O caused the detection reported by the message D at time t. The vehicle index i is incremented in step S43 and if it is determined in step S44 that i is less than the number of vehicles, the steps S41 to S43 are repeated. Otherwise in step S45 it is determined which vehicle caused the detection reported by the message D at time t with the highest probability. In step S46 the index of that vehicle is returned as the result if the method.
- A method to estimate (update the present estimation of) the state of a vehicle is now described in more detail with reference to
Figure 13 . - In step S60 the messages D1,...,Dn associated with vehicle O are selected.
- In step S61 a probability density function is constructed on the basis of the associated messages D1,...,Dn.
- In step S62 the current state S0 and time to for vehicle O is retrieved from the vehicle database.
- In step S63 it is determined whether the time for which the state S of the vehicle O has to be determined is greater than the time to associated with the current state S0.
- If that is the case, the state S (determined by the estimation method) is the state update of S0 to t, performed in step S65. If that is not the case, then the message D relates to a detection preceding the detection that resulted in the earlier estimation for state S0. In that case the state S0 is updated using the detection D by the state estimation method in step S64
- In the claims the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single component or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different claims does not indicate that a combination of these measures cannot be used to advantage. Any reference signs in the claims should not be construed as limiting the scope. Further, unless expressly stated to the contrary, "or" refers to an inclusive or and not to an exclusive or. For example, a condition A or B is satisfied by any one of the following: A is true (or present) and B is false (or not present), A is false (or not present) and B is true (or present), and both A and B are true (or present).
- More details relevant for the present invention are described in the following
-
- A1: Estimation and association for multiple target tracking based on spatially, distributed detections
- A2: On Event Based State Estimation
- Summary. In this paper we consider the multiple object tracking problem with event-based observations. For that we predefine a number detection points which are spatially distributed along the road. Whenever the edge of an object crosses one of the detection points, the position of that detection point together with the time of the event are received by our tracking algorithm. We assume that objects can cover multiple detection points and propose a method to estimate the object's position and orientation from these detections using the shape of the object. Beside that another method is designed which associates newly received detections with a known object. The objects are tracked with an event-based state-estimator that uses the estimated position and orientation, although its design is out of the scope of this paper. Finally our tracking algorithm is critically assessed in a simulation of a parking lot.
- In multiple target tracking [1-3] one aims to track all the objects/targets, which are moving in a certain area. Three basic problems arise from tracking objects. The first one is how to measure the object's position. The second one is to associate a certain measurement with its correct object and the third one is a state-estimator to keep track of all the objects. This paper considers the first 2 issues when objects are not measured but detected.
- Consider a system in which objects are detected when they cross a predefined 'detection' point. These detectors are triggered by the event that the object's edge crosses its position. However, they cannot distinguish between the objects. This paper describes a method in which a new detection is associated with the object that most probable generated it. Also, a method is described which estimates the position and orientation of the object given the observations in position and time due to the detections. Other examples in which sensor-data is generated due to an event can be found in [4, 5].
- This paper is organized as follows.
Section 2 defines background knowledge such as the notation of (object) variables and functions that are used throughout this paper. After that the problem is formulated insection 3 together with existing methods.Section 5 describes the approach which is taken in the design. A more detailed description of the estimation and associated is presented inSection 5 and 6 respectively. Finally both methods are tested in a small application example presented in Section 6 and conclusions are drawn in section 7. But let's start with the background information. - In order to be clear about notations and variables this section describes those that can be found throughout this paper.
- R defines the set of real numbers whereas the set R+ defines the non-negative real numbers. R xy defines the set spanned by the vectors ex and ey , the point p := x·ex + y·ey is shortly denoted as p = (x,y) T. The set Z defines the integer values and Z+ defines the set of non-negative integer numbers. The variable 0 is used either as null, the null-vector or the null-matrix. Its size will become clear from the context.
-
- The matrix A(t 2 -t 1 )∈ R a×b depends on the difference between two time instants t2 > t 1 and is shortly denotes as A t
2 -t1 - The transpose, inverse and determinant of a matrix A∈ R n×n are denoted as AT , A -1 and |A| respectively.
- Let us define the probability of the random vector x∈R n as the scalar Pr(x) ∈ {0,1} and the conditional probability of x given the vector u ∈ R m the scalar Pr(x I u) ∈ {0,1}. The probability density function (PDF), as defined in [6] section B2, of the vector x∈R n is denoted as p(x) and the conditional PDF of x given u ∈ R q is denoted as p(x|u). The expectation and covariance of x are denoted as E[x] and cov(x) respectively. The conditional expectation of x given a vector u is denoted as E[x|u]. The definitions of E[x], E[x|u] and cov(x) can be found in [6] sections B4 and B7.
-
- If p(x) = G(x,u,P), then by definition it holds that E[x] = u and cov(x) = P.
-
- Assume there exist an object which is moving in the 3D world space. This object is observed with, for example, a camera or sensors in the road. Meaning that the object is projected to a 2D space, i.e. R xy . If we assume that the shape of the projected object is constant and known, then we can draw a smallest, rectangular box around the object. For the box we define a position-vector o = (x, y) T ∈ R xy , equal to the center of the box, and an orientation-vector θ ∈ R. In the case of o = 0 and θ = 0 the corners of this box, as shown in
Figure 14 , are defined in the set C0 : -
- Beside the positions o and θ each object also has a certain shape or geometry which covers a certain set of positions in R xy , i.e. the grey area of
Figure 14 . This closed set is denoted with S ⊂ R xy and is defined as the union of the open set of the object's body SB ⊂ R xy and the closed set of the object's edge SE ⊂ R xy , i.e. S := SB ∪SE. The set S is approximated by a set of sampled position-vectors A = [λ 1,λ 2,···,λK ], with λi ∈ R xy . To define the vectors λ i we equidistant sample the rectangular box defined by C0 using a grid with a distance r. Each λ i is a grid point within the set S as graphically depicted inFigure 15 . -
- Next the problem is formulated using this background knowledge.
- A total of E objects are observed within the set R xy . The vectors oi = (xi, yi ) T and θi are the ith object's position- and rotation-vector respectively. Ti represents the ith object's rotation-matrix dependent on θi. The dynamical process of object i with state-vector si , process-noise wi and measurement-vector mi is defined with the following state-space model:
-
- The objects are observed in R xy by a camera or a network of sensors. For that M 'detection' points are marked within R xy and collected in the set D ⊂ R xy . The position of a detection point is denoted as d ∈ D. The kth detection of the system generates the observation vector
- However, the system does not know which object was detected for it can be any object. As a result the system will not generate
- From equations (9) and (10) we conclude that
-
Figure 16 shows an example of object i which is detected by multiple detection points. The covariance ε of each detection point is also indicated. - The sampling method of the observation vectors z 0:k is a form of event sampling [4, 5, 7]. For a new observation vector is sampled whenever an event, i.e. object detection, takes place. With these event samples all N objects are to be tracked. To accomplish that three methods are needed. The first one is the association of the new observation-vector zk to an object i and therefore denote it with
- Present association methods are: Gating and Nearest Neighbor (NN) [2], (Joint) Probabilistic Data Association ((J)DPA) [2, 8], Multiple Hypothesis Tracker (MHT) [9] and the MCMCDA [10]. Although these can be transformed for associating the event samples z 0:k , this paper will show that the estimation of
-
- The set
-
- 1. The first step is to position the object on each detection point dn and mirror its set S into the set On, as shown in
Figure 17 for a single detection. This way we transform the points that are covered by the object, into possible vectors of the object's position - 2. The second step, graphically depicted in
Figure 18 , is to turn all sets On simultaneously around their detection point dn. This way, each possible orientation -
- Estimation of the measurement-vector
- Although the measurement vector is defined as m = (o, θ) T, with o ∈ R xy and θ ∈ R, the detection point at time-step n are defined as dn ∈ R xy . Meaning that the objects orientation is not directly. However, because every observation vector zn ∈ Z detects the object for one and the same θ, the PDF p(m|Z) is approximated by sampling in θ, i.e.:
- The main aspect of equation (17) is to determine p(o|Z,θ). To do that we define the set On (θ) ∈ R xy to be equal to all possible object positions o, given that the object is detected at position dn ∈ zn (∈ Z) and that the object's rotation is equal to θ. The determination of On (θ) ∈ R xy is presented in the n the next section.
-
- Equation (19) is graphically explained in
Figure 19 for two different values of θ and N={1,2}. -
-
- With (21) both p(m|Z) is calculated according to (6). The rest of this section is divided into two parts. The first part derives the probability function based on a single detection, i.e. f(o|zn,θ). While the second part derives the probability function based on a multiple detections, i.e. g(o|Z,θ).
- In order to derive f(o|z n,θ) we will use the set Λ, defined in 2.3, which contains the sampled positions λi that are covered by the object if o = 0 and θ = 0. Notice that if the object covers the origin, i.e. (x, y) T = 0, then the possible values of the object position o are given by the set - Λ. This is graphically depicted in
Figure 20 (left). From that we can conclude that if the object covers the detection point dn , given a certain orientation θ and rotation-matrix T, the sampled set Λ can be transformed into a sampled set of On , denoted with O n : -
Figure 20 (right) graphically depicts the determination of Ôn from the set Λ for a given θ and detection point dn . -
- The approximation of (18) assumes that the object is detected exactly at dn. In
Section 4 we stated that the detection can be a bit of a detection point. The PDF that the object is detected at position υ ∈ R xy given the detection point dn is defined in (15). Inserting this uncertainty into (18) results in the final f (o | zn , θ): - Equation (19) is solved with the following Proposition and the fact that G(x,a+b,C) = G(x-b,a,C) :
-
Proposition 1. Let there exist two Gaussian functions of the random vectors x ∈ R n and m ∈ R q and the matrix Γ ∈ R q×n ; G(x,u,U) and G(m,Γx,M). Then they have the following property: -
- From f(o|z n,θ) based on a single detection, the next step to multiple detections, i.e. g(o|Z,θ), is taken.
-
- If N contains m elements, then calculating equation (22) would result in Km products of m Gaussian functions and sum them afterwards. This would take too much processing power if m is large. That is why equation (22) is calculated differently.
- Instead of using all detection points dn we will use a subset of them. The derivation of this subset is graphically depicted in
Figure 21 for N = {1,2}. For that consider the rectangular set C0 ∈ R xy of Section 2.3 defined by its corners [c 1,c 2,c 3,c 4]. For each detection point dn we define the set C n (θ) ⊂ R xy with corner-points [c n1 (θ),c n2 (θ),cn3 (θ),c n4 (θ)] defined as: -
- Meaning that each detection point dn defines a rectangular set denoted with C n (θ) dependent on rotation θ. The intersection of all these rectangular sets is defined with the set C N (θ).
- In the beginning of this section we defined two different sets shown in
Figure 17 and 18 . The first set, On (θ),shown inFigure 17 defines all possible objet positions o based on a single detection at dn . The second set, i.e. ON (θ),shown inFigure 18 , defines all possible object positions o based on all detections at dn , ∀n ∈ N. Notice that as a result On (θ) ⊂ C n (θ) and ON (θ) ⊂ C N (θ). Meaning that only within the set CN (θ) all the functions f (o | zn,θ) have an overlapping area in which they are 1. Outside CN (θ) there is always at least one f(o|z n, θ) which is 0 and therefore makes g(o|Z,θ) outside C N (θ) equal to 0. Therefore g(o|Z,θ) of (22) can be approximated by taking only those Gaussians of the functions f (o | z n ,θ) into account of which their mean, i.e. -
- The calculation of (26) is done by applying the following two propositions. The first one, i.e.
Proposition 2, shows how to rewrite a product of a summation of Gaussians into a summation of a product of Gaussians. The second one, i.e.Proposition 3, proofs that a product of Gaussians results in a single Gaussian. -
Proposition 2. The product of α summation of Gaussians can be written into α summation of a product of Gaussian: -
Proposition 3. The product of Gaussians is again a Gaussian:Section 10. -
-
-
- As was mentioned in the problem formulation, the PDF p(m|Z) also gives us the probability that a new observation vector is generated by an certain object i. This is discussed in the next section.
- The total probability that a new observation vector zk is generated by object i is equal to the total probability of the measurement-vector
- The variables γi and Ki are equal to γ and K respectively, which define the approximation of the function
- Now that both the measurement estimation as well as the detection association are designed, both are tested in a multiple object tracking application.
- As an application example we take a parking lot of 50 by 50 meters with a network of wireless sensors distributed randomly along the road's surface. Each sensor can detect a crossing vehicle. A total of 2500 sensors was used resulting in a density of one sensor per square meter. The vehicles are all assumed rectangular shaped objects with a length of 5 meters and a width of 2 meters. A total of 4 vehicles manoeuvre within the parking lot and are tracked using a data-associator followed with an event state-estimator.
- The simulation case is made such that it contains two interesting situation. One in which two vehicles cross each other in parallel and one where two vehicles cross perpendicular. For comparison the objects are tracked using two different association methods. The first one is a combination of Gating and detection association of 6. The second one is a combination of Gating and Nearest Neighbor.
- The result of the detection associator (DA) for both crossings is shown in
Figure 22 while the result of the Nearest Neighbor (NN) associator is shown inFigure 23 . In both results the real object is plotted in a thick, solid line while its estimated one is plotted in a thin, solid line. The associated detections of each object are given with a symbol which is different for each object; '□' if associated withvehicle 1, '○' if associated withvehicle 2, '∇' if associated withvehicle 3 and '*' if associated withvehicle 4. -
Figure 22 shows with the DA all detections were correctly associated to the one object, while If NN is used as an association method,we see that a lot of incorrect associated detections. Therefore we can concluded that using the detection association of 7 results in less estimation-error compared to NN. - A second simulation is done to compare the percentage of incorrect associated detections. Again for the both DA as well as NN only now 4 different amount of detection points were used: 3000, 2500, 2000 and 1500. This table shows that the detection association has a better performance compared to Nearest Neighbor.
Table 1: Percentage of incorrect association amount of detection points DA NN 3000 0 % 4.5 % 2500 0 % 5.6 % 2000 0 % 7.8 % 1500 0 % 2.2 % - This paper presents a method for estimating the position- and rotation-vector of objects from spatially, distributed detections of that object. Each detection is generated at the event that the edge of an object crosses a detection point. From the estimation method a detection associator is also designed. This association method calculates the probability that a new detection was generated by an object i.
- An example of a parking lot shows that the detection association method has no incorrect associated detections in the case that two vehicles cross each other both in parallel as well as orthogonal. If the association method of Nearest Neighbor was used, a large amount of incorrect associated detections were noticed, resulting in a higher state-estimation error.
- The data-assimilation can be further improved with two adjustments. The first one is replacing the set S with SE only at the time-instants that the observation vector is received. The second improvement is to take the detection points that have not detected anything also in account.
-
- 1. Poore and S. Gadaleta, "Some assignment problems arising from multiple target tracking," Mathematical and Computer Modelling, vol. 43, pp. 1074--1091, 2006.
- 2. Y. Bar-Shalom and R. Li, Multitarget-Multisensor Tracking: Principles and Techniques. YBS, 1995.
- 3. S. Blackman and R. Popoli, Design and Analysis of Modern Tracking Systems. Artech House, Norwood, MA, 1999.
- 4. V. Nguyen. and Y. Suh, "Improving estimation performance in Networked Control Systems applying the Send-on-delta transmission method," Sensors, vol. 7, pp.2128-2138, 2007.
- 5. K. Åström and B. Bernhardsson, "Comparison of Riemann and Lebesque sampling for first order stochastic systems," in 41st IEEE Conference on Decision and Control, Las Vegas, Nevada, USA, 2002.
- 6. N. Johnson, S. Kotz, and A Kemp, Univariate discrete distributions. JOHN WILEY and SONS, Inc, 1992.
- 7. M. Miskowicz, "Send-on-delta concept: an event-based data-reporting strategy," Sensors, vol. 6, pp. 49-63, 2006.
- 8. R. Karlsson and F. Gustafsson, "Monte Carlo data association for multiple target tracking," IEE International Seminar on Target Tracking: Algorithms and Applications, 2001.
- 9. Ristic, S. arulampalam, and N. Gordon, Beyond the Kalman filter: Particle filter for tracking applications. 2002.
- 10. Songhwai, S. Sastry, and L. Schenato, "A Hierarchical Multiple-Target Tracking Algorithm for Sensor Networks," in Proc. of the 2005 IEEE International Conference on Robotics and Automation, Barcelona, Spain, 2005.
- 11. K. Mardia, J. Kent, and J. Bibby, Mutlivariate analysis. Academic press, Inc. London, 1979.
- 12. Kelly, Introduction to probability. Macmillan Publishing Company, 1994.
- 13. H. Durant-Whyte, B. Rao, and H. Hu, "Towards a fully decentralized architecture for multi-sensor data fusion," in 1990 IEEE Int. Conf. on Robotics and Automation, Cincinnati, Ohio, USA, 1990, pp. 1331-1336.
-
-
-
-
-
-
-
- Summary. To reduce the amount of data transfer in networked control systems and wireless sensor networks, measurements are usually taken only when an event occurs, rather that at each synchronous sampling instant. However, this complicates estimation and control problems considerably. The goal of this paper is to develop a state estimation algorithm that can successfully cope with event based measurements. Firstly, we propose a general methodology for defining event based sampling. Secondly, we develop a state estimator with a hybrid update, i.e. when an event occurs the estimated state is updated using measurements; otherwise the update is based on the knowledge that the monitored variable is within a bounded set used to define the event. A sum of Gaussians approach is employed to obtain a computationally tractable algorithm.
- Different methods for state estimation have been introduced during the last decades. Each method is specialized in the type of process, the type of noise or the type of system architecture. In this paper we focus on the design of a state estimator that can efficiently cope with event based sampling. By even sampling we mean that measurements are generated only when an a priori defined event occurs in the data monitored by sensors. Such an effective estimator is very much needed in both networked control systems and wireless sensor networks (WSNs) [1]. Especially in WSNs, where the limiting resource is energy, data transfer and processing power must be minimized. The existing estimators that could be used in this framework are discussed in
Section 4. For related research on event based control, the interested reader is referred to the recent works [2], [3]. - The contribution of this paper is twofold. Firstly, we introduce a general mathematical description of event based sampling. We assume that the estimator does not know when new measurements are available, which usually results in unbounded eigenvalues of its error-covariance matrix. To obtain an estimator with a bounded error-covariance matrix, we develop an estimation algorithm with hybrid update, which is the second main contribution. The developed event based estimator is updated both when an event occurs, with received measurements, as well as at sampling instants synchronous in time. Then the update is based on the knowledge that the monitored variable is within a bounded set used to define the event. In order to meet the requirements of a low processing power, the proposed state estimator is based on the Gaussian sum filter [4, 5], which is known for its computational tractability.
- R defines the set of real numbers whereas the set R+ defines the non-negative real numbers. The set Z defines the integer numbers and Z+ defines the set of non-negative integer numbers. The
notation 0 is used to denote either the null-vector or the null-matrix. Its size will become clear from the context. - A vector x(t)∈ R n is defined to depend on time t ∈ R and is sampled using some sampling method. Two different sampling methods are discussed. The first one is time sampling in which samples are generated whenever time t equals some predefined value. This is either synchronous in time or asynchronous. In the synchronous case the time between two samples is constant and defined as ts ∈ R+. If the time t at sampling instant ka ∈ Z+ is defined as tk
a , with t0 a := 0, we define:e , with t 0 e := 0, we define: - A transition-matrix A t
2 -t1 ∈ R a×b is defined to relate the vector u(t 1)∈ R b to a vector x(t 2) ∈ R a as follows: x(t 2) = A t2 -t1 u(t 1). - The transpose, inverse and determinant of a matrix A∈ R n×n are denoted as AT , A -1 and |A| respectively. The ith and maximum eigenvalue of a square matrix A are denoted as λi (A) and λmax (A) respectively. Given that A∈ R n×n and B∈ R n×n are positive definite, denoted with A ≻ 0 and B ≻ 0, then A ≻ B denotes A - B ≻ 0. A±0 denotes A is positive semi-definite.
- The probability density function (PDF), as defined in [6] section B2, of the vector x∈ R n is denoted with p(x) and the conditional PDF of x given u ∈ R q is denoted as p(x|u). The expectation and covariance of x are denoted as E[x] and cov(x) respectively. The conditional expectation of x given u is denoted as E[x|u]. The definitions of E[x], E[x|u] and cov(x) can be found in [6] sections B4 and B7.
-
- If p(x) = G(x,u,P), then by definition it holds that E[x] = u and cov(x) = P.
-
-
- Many different methods for sampling a vector y(t)∈ R q can be found in literature. The one mostly used is time sampling in which the
a := t ka -1 + τ ka -1 for some τ ka -1 ∈ R+. Recall that if y(t) is sampled at ta it is denoted as yka . This method is formalized by defining the observation vectora (z ka -1) ⊂ R containing all the values that t can take between t ka -1 and t ka -1+τ ka -1, i.e.: -
- In the case of synchronous time sampling τk
a = ts , ∀ ka ∈ Z+, which is graphically depicted inFigure 24 . Notice that with time sampling, the present time t specifies when samples of y(t) are taken, but time t itself is independent of y(t) . As a result y(t) in between the two samples can have any value within R q . Recently, asynchronous sampling methods have emerged, such as, for example "Send-on-Delta" [9, 10] and "Integral sampling" [11]. Opposed to time sampling, these sampling methods are not controlled by time t, but by y(t) itself. - Next, we present a general definition of event based sampling, which recovers the above mentioned asynchronous methods, for a particular choice of ingredients. Let us define the observation vector at sampling instant ke-1 as
e (z ke-1,t) ⊂ R q+1, which depends on both z ke -1 and t. In line with time sampling the next event instant, i.e. ke , takes place whenever y(t) leaves the set Hke (z ke-1,t) as shown inFigure 25 for q = 2. Therefore z ke is defined as:e (z ke -1,t) depends on the actual sampling method. As an example Hke (z ke-1,t) is derived for the method "Send-on-Delta", with y(t) ∈ R. In this case the event instant ke occurs whenever |y(t) - y ke -1| exceeds a predefined level Δ, seeFigure 26 , which results in Hke (z ke-1,t)={y∈ R|-Δ< y-y ke -1 <Δ}. - In event sampling, a well designed Hk
e (z ke-1,t) should contain the set of all possible values that y(t) can take in between the event instants ke-1 and ke. Meaning that if t ke -1≤t<tke , then y(t)∈ Hke (z ke -1,t). A sufficient condition is that y ke -1∈ Hke (z ke -1 ,t), which for "Send-on-Delta" results in y(t)∈ [y ke -1-Δ,y e-1 +Δ] for all t ke -1≤t<tke . - Assume a perturbed, dynamical system with state-vector x(t) ∈ R n , process-noise w(t) ∈ R m , measurement-vector y(t) ∈ R q and measurement-noise v(t) ∈ R q . This process is described by a state-space model with Aτ ∈ R n×n , Bτ ∈ R n×m and C ∈ R q×n . An event sampling method is used to sample y(t). The model of this process becomes:
- The state vector x(t) of this system is to be estimated from the observation vectors z0 e : k
e . Notice that the estimated states are usually required at all synchronous time samples ka , with ts = t ka -t ka-1, e.g., as input to a controller that runs synchronously in time. As such our goal is to construct an event-based state-estimator (EBSE) that provides an estimate of x(t) not only at the event instants tke but also at the sampling instants tka . Therefore, we define a new set of sampling instants tn as the combination of sampling instants due to event sampling, i.e. ke , and time sampling, i.e. ka : - The estimator calculates the PDF of the state-vector xn given all the observations until tn . This results in a hybrid state-estimator, for at time tn an event can either occur or not, which further implies that measurement data is received or not, respectively. In both cases the estimated state must be updated (not predicted) with all information until tn . Therefore, depending on tn a different PDF must be calculated, i.e.: if
-
- The PDFs of (9) can be described as the Gaussian G(xn, x n | n ,P n|n ). The square root of the eigenvalues of P n|n , i.e.
Figure 27 for the 1D case andFigure 29 for a 2D case, in a top view. The smaller the eigenvalues λi (P n|n ) are, the smaller the estimation-error is. - As such, the problem of interest in this paper is to construct a state-estimator suitable for the general event sampling method introduced in
Section 3 and which is computationally tractable. Furthermore, it is desirable to guarantee that P n|n has bounded eigenvalues for all n. - Existing state estimators can be divided into two categories. The first one contains estimators based on time sampling: the (a)synchronous Kalman filter [12, 13] (linear process, Gaussian PDF), the Particle filter [14] and the Gaussian sum filter [4, 5] (nonlinear process, non-Gaussian PDF). These estimators cannot be directly employed in event based sampling as if no new observation vector zk
e is received, then tn-tke → ∞ and λi (P n|ke -1)→∞. The second category contains estimators based on event sampling. In fact, to the best of our knowledge, only the method proposed in [15] fits this category. However, this EBSE is only applicable in the case of "Send-on-Delta" event sampling and it requires that any PDF is approximated as a single Gaussian function. Moreover, the asymptotic property of P n|n is not investigated in [15]. - In the next section we propose a novel event-based state-estimator, suitable for any event sampling method, together with a preliminary result on asymptotic analysis.
-
-
-
-
- The calculation of p(xn |y 0:n ∈ Y 0:n ) is done in three steps: 1. Assimilate p(yn ∈ Yn |xn ) for both tn = tk
e and tn =tka . 2. Calculate p(xn|y 0:n ∈ Y 0:n ) as a summation of N Gaussians. 3. Approximate p(xn |y0:n ∈ Y 0:n ) as a single Gaussian function. The reason for this last step is to design an algorithm in which p(xn |y 0:n ∈ Y 0:n ) is described by a finite set of Gaussians and therefore attain computational tractability. Notice that (13) gives a unified description of the hybrid state-estimator, which makes an asymptotic analysis of the EBSE possible, as it will be shown later in this section. -
-
- Substitution of (16) into (15) gives that p(yn ∈ Yn |xn ) =G(yk
e ,Cxn ,V) if tn =tke . However, if tn =tka then p(yn ∈ Yn |xn ) equalse and therefore on the actual event sampling method that is employed. In order to have a unified expression of p(yn ∈ Yn |xn ) for both types of tn , independent of the event sampling method,Figure 29 for yn ∈ R2. The interested reader is referred to [4 for more details. -
-
Proposition 1. [12, 14] Let there exist two Gaussians of random vectors x ∈ R n and m ∈ R q , with Γ ∈ R q×n : G(m,Γx,M) and G(x,u,U). Then they satisfy:e and tn =tka , for any event sampling method results in:e the variables of (22) are: N=1,a the variables depend on
In "Send-on-Delta"', for certain N, the approximation of - With the result of (22), p(xn |y 0:n ∈ Y 0:n ) can also be expressed as a sum of N Gaussians.
-
-
-
-
- The third step is to approximate (27) as a single Gaussian to retrieve a computationally tractable algorithm. For if both p(x n-1|y 0:n-1 ∈ Y0:n-1) and p(yn ∈ Yn |xn ) are approximated using N Gaussians, the estimate of xn in (27) is described with Mn Gaussians. The value of Mn equals M n-1 N, meaning that Mn increases after each sample instant and with it also the processing demand of the EBSE increases.
-
- The expectation an covariance of (27), equal to x n|n and P n|n of (28), can be derived from the corresponding definitions. Notice that because the designed EBSE is based on the equations of the Kalman filter, the condition of computational tractability is met.
- In this section we investigate the asymptotic analysis of the error-covariance matrix of the developed EBSE. By this we mean that we analyze limn→∞P n|n, which for convenience is denoted as P ∞. Note that for the classical Kalman filter (KF) [12] such an analysis is already available. However, for any other type of estimator asymptotic analysis remains a very challenging problem, which is why in most cases it is not even considered.
-
- In [18, 19] it is proven that if the eigenvalues of At
s are within the unit circle and (Ats , C) is observable, then P ∞ = PK. The matrix PK equals the solution of:Section 4 this is a valid indication of an estimator's performance. - The main result of this section is obtained under the standing assumption that
n -matrix are within the unit-circle and (Aτn ,C) is an observable pair. The following technical Lemmas will be of use. -
Lemma 1. Given the process model (7) and covariance matrices P ≻ 0 and Q ≻ 0, then for any 0 < τ 1 ≤ τ 2 we have that -
Lemma 2. Let any square matrices V 1°V 2 and W1°W 2 with V 1±0 and W1±0 be given. Suppose that the matrices U 1 and U 2 are defined as -
-
Theorem 1. Suppose that the EBSE, as presented in Section 6, approximates - In this section we illustrate the effectiveness of the developed EBSE in terms of state-estimation error, sampling efficiency and computational tractability. The case study is a 1D object-tracking system. The states x(t) of the object are position and speed while the measurement vector y(t) is position. The process-noise w(t) represents the object's acceleration. Then given a maximum acceleration of 0.5[m/s 2] its corresponding Q, according to [21], equals 0.02. Therefore the model as presented in (7) yields
Figure 30 together with the object's position and speed. The sampling time is ts = 0.1 and the measurement-noise covariance is V = 0.1.10-3. - Three different estimators are tested. The first two estimators are the EBSE and the asynchronous Kalman filter (AKF) of [13]. For simplicity, in both estimators we used the "Send-on-Delta" method with Δ=0.1[m]. For the EBSE we approximated
e . The states at tka are calculated by applying the prediction-step of (14b). The third estimator is based on the quantized Kalman filter (QKF) introduced in [21] that uses synchronous time sampling of yka . The QKF can deal with quantized data, which also results in less data transfer, and therefore can be considered as an alternative to EBSE. In the QKFy ka is the quantized version of yka with quantization level 0.1, which corresponds to the "Send-on-Delta" method. Hence, a comparison can be made. - In
Figure 31 and Figure 32 the state estimation-error of the three estimators is plotted. They show that the QKF estimates the position of the object with the least error. However, its error in speed is worse compared to the EBSE. Further, the plot of the AKF clearly shows that prediction of the state-estimates gives a significant growth in estimation-error when the time between the event sampling-instants increases (t > 4). - Beside estimation error, sampling efficiency η is also important due to the increased interest in WSNs. For these systems communication is expensive and one aims to have the least data transfer. We define η ∈ R+ as
e ory ka was done. Notice that if η < 1 the estimation error decreased after an update, if η > 1 the error increased and if η = 1 the error remained the same. For the EBSE i=ke with i-1 equal to ke -1 or ka -1. For the AKF i=ke with i-1=ke -1. For the QKF i=ka and i-1=ka -1.Figure 33 shows that for the EBSE η < 1 at all instants n . The AKF has one instant, t=3.4, at which η >1. In case of the QKF the error sometimes decreases but it can also increase considerably after an update. Also notice that η of the QKF converges to 1. Meaning that for t > 5.5 the estimation error does not change after an update and new samples are mostly used to bound λi (P ka |ka ). The EBSE has the same property, although for this method the last sample was received at t=4.9. - The last aspect on which the three estimators are compared is the total amount of processing time which was needed to estimate all state-vectors. For the EBSE, both xk
e and xka were estimated and it took 0.094 seconds. The AKF estimated xke and predicted xka in a total time of 0.016 seconds and the QKF estimated xka and its total processing time equaled 0.022 seconds. This means that although the EBSE results in the most processing time, it is computationally comparable to the AKF and QKF, while it provides an estimation-error similar to the QKF, but with significantly less data transmission. As such, it is most suited for usage in networks in general and WSNs in particular. - In this paper a general event-based state-estimator was presented. The distinguishing feature of the proposed EBSE is that estimation of the states is performed at two different type of time instants, i.e. at event instants tk
e , when measurement data is used for update, and at synchronous time sampling tka , when no measurement is received, but an update is performed based on the knowledge that the monitored variable lies within a set used to define the event. As a result, it could be proven that, under certain assumptions, for the error-covariance matrix of the EBSE it holds that λmax (P ∞) <∞, even in the situation when no new observation zke is received anymore. Its effectiveness for usage in WSNs has been demonstrated on an application example. -
- 1. F. Akyildiz, W. Su, Y. Sankarasubramnniam, and E. Cayirci, "Wireless Sensor Networks:a survey," Elsevier, Computer Networks, vol. 38, pp. 393-422, 2002.
- 2. E. Johannesson, T. Henningsson, and A. Cervin, "Sporadic control of first-order linear stochastic systems," in Hybrid Systems: Computation and Control, ser. Lecture Notes in Computer Science, vol. 4416. Pisa, Italy: Springer Verlag, 2007, pp. 301-314.
- 3. W. P. M. H. Heemels, J. H. Sandee, and P. P. J. van den Bosch, "Analysis of event-driven controllers for linear systems," International Journal of Control, vol. 81, no. 4, 2008.
- 4. H. W. Sorenson and D. L. Alspach, "Recursive Bayesian estimation using Gaussian sums," Automatica, vol. 7, pp. 465-79, 1971.
- 5. J. H. Kotecha and P. M. Djuric, "Gaussian sum particle filtering," IEEE Transaction Signal Processing, vol. 51, no. 10, pp. 2602-2612,2003.
- 6. N. L. Johnson, S. Kotz, and A, W. Kemp, Univariate discrete distributions, John Wiley and Sons, 1992.
- 7. L. Aggoun and R. Elliot, Measure Theory and Filtering. Cambridge University Press, 2004.
- 8. H. L. Lebesque, "Integrale, longueur, aire," Ph.D. dissertation, University of Nancy, 1902.
- 9. K. J. Astrom and B. M. Bernhardsson, "Comparison of Riemann and Lebesque sampling for first order stochastic systems," in 41 st IEEE Conf, on Dec, and Contr., Las Vegas, USA, 2002.
- 10. M. Miskowicz, "Send-on-delta concept: an event-based data-reporting strategy," Sensors, vol. 6, pp. 49-63, 2006.
- 11. "Asymptotic Effectiveness of the Event-Based Sampling according to the Integral Criterion," Sensors, vol. 7, pp. 16-37, 2007.
- 12. R. E. Kalman, "A new approach to linear filtering and prediction problems," Transaction of the ASME Journal of Basic Engineering, vol. 82, no. D, pp. 35-42, 1960.
- 13. M. Mallick, S. Coraluppi, and C. Carthel, "Advances in Asynchronous and Decentralized Estimation" in In Proceeding of the 2001 Aerospace, Conference, Big Sky, MT, USA, 2001.
- 14. B. Ristic, S. Arulampalam, and N. Gordon, Beyond the Kalman filter: Particle filter for tracking applications. Boston, Massachusetts: Artech House, 2004.
- 15. V. H. Nguyen and Y. S. Sun, "Improving estimation performance in Networked Control Systems applying the Send-on-delta transmission method," Sensors, vol. 7, pp. 2128-2138, 2007.
- 16. K. V. Mardia, J. T. Kent, and J. M, Bibby, Mutlivariate analysis. Academic press, London, 1979.
- 17. D. C. Montgomery and G. C, Runger, Applied Statistics and Probability for Engineers. John Wiley and Sons, 2007.
- 18. M. L. J. Hautus, "Controllability and observability conditions of linear autonomous systems," in Indagationes Mathemathicae, vol. 32, 1972, pp. 448-455.
- 19. V. Balakrishnan, Kalman Filtering Theory. Optimization Software, New York, 1987.
- 20. D. S. Bernstein, Matrix Mathematics. Princeton University Press, 2005.
- 21. R. E. Curry, Estimation and control with quantized measurements. Boston, Massachusetts: MTT Press, 1970.
- 22. G. F. Franklin, J. D. Powel, and A. Emami-Naeini, Feedback control of dynamic systems. Addison-Wesley, 1995.
-
-
- Under the hypothesis, for the proposed EBSE, P n|n of (6.3), with τn :=tn-t n-1 and
V . Notice that for these estimators we have that τ n ≤ ts and Rn °R, for all n. Let the EBSE and the KF start with the same initial covariance matrix P 0. The first step of induction is to prove that -
-
-
Claims (14)
- Vehicle tracking system comprising- a plurality of sensor nodes (10) that each provide a message (D) indicative for an occupancy status of a detection area of a vehicle infrastructure (80) monitored by said sensor node,- a message interpreter (MI) including a vehicle database facility (32, 34) with state information of vehicles present at the vehicle infrastructure, and a database updating facility (DBU) for updating the vehicle database facility (32, 34) on the basis of messages (D) provided by the sensor nodes,characterized in that multiple sensor nodes (10) are arranged in the vehicle infrastructure (80) at a density of at least 0.2 per square meter.
- The vehicle tracking system according to claim 1, wherein said sensor nodes (10) are arranged in the vehicle infrastructure (80) at a density of at least 0.6 per square meter.
- The vehicle tracking system according to claim 1 or 2, comprising a plurality of system modules (MD1, MD2, MD3), each module comprising a respective subset of the plurality of sensor nodes (10) for monitoring a respective section (80A, 80B, 80C, 80D) of the vehicle infrastructure and a respective message interpreter (MI), the vehicle tracking system further having a communication facility (60) for enabling system modules of mutually neighboring sections to exchange state information.
- Vehicle tracking system according to claim 1 or 2, wherein the database updating facility (DBU) comprises- an association facility (40) for associating the messages (D) provided by the sensor nodes with the state information present in the vehicle data base facility (32, 34),- a state updating facility (50) for updating the state information on the basis of the messages (D) associated therewith.
- Vehicle tracking system according to claim 1 or 2, wherein the sensor nodes (10) provide the messages (D) at an event basis.
- Vehicle tracking system according to claim 1 or 2, wherein the sensor nodes (10) are provided with a wireless transmission facility (16) for wirelessly transmitting the message (D), and wherein the message interpreter (40) comprises a wireless reception facility (20) for receiving the message (D).
- The vehicle tracking system according to claim 1 or 2, wherein the sensor nodes (10) are randomly distributed over the vehicle infrastructure (80).
- The vehicle tracking system according to claim 1 or 2, wherein the state updating facility (50) is arranged for updating the state information on the basis of event-based messages (D) and on the basis of messages sampled synchronous in time.
- Vehicle infrastructure (80) provided with a vehicle tracking system according to one of the previous claims.
- Method for tracking vehicles at an infrastructure using the vehicle tracking system according to one of claims 1 to 8, the method comprisinga) providing a plurality of sensor nodes in the vehicle infrastructure at a density of at least 0.2 per square meter, the sensor nodes each monitoring a detection area of the vehicle infrastructure,b) providing a message indicative for an occupancy status of their detection area,c) storing state information of vehicles present at the vehicle infrastructure,d) receiving the message and updating said stored information on the basis of the message.
- Method according to claim 10, wherein the density of the sensor nodes is at least 0.6 per square meter.
- Method according to claim 10 or 11, wherein the step of updating said stored information comprises,d) associating the message with state information present in the vehicle data base facility,e) updating the state information associated in step d) on the basis of said message.
- Method according to claim 10 or 11, wherein the steps a) to c) are independently performed for mutually non coinciding sections of the vehicle infrastructure, the method further comprising the step of exchanging state information.
- Method according to claim 12 wherein the step of associating comprises- initializing (S40) a vehicle index (i),- retrieving (S41) the current state known for the vehicle with that index from a vehicle database facility,- determining (S42) a probability that the vehicle with that index caused the detection reported by the message D,- incrementing (S43) the vehicle index,- determining (S44) whether the vehicle index is less than the number of vehicles,- if the outcome of the determination is positive repeating steps S41 to S43 with the incremented vehicle index, and- if the outcome of the determination is negative, determining (S45) which vehicle caused the detection reported by the message D with the highest probability.- returning (S46) the index of that vehicle identified in step S45.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP09771423.2A EP2374117B1 (en) | 2008-12-12 | 2009-12-11 | Vehicle tracking system, vehicle infrastructure provided with vehicle tracking system and method for tracking vehicles |
Applications Claiming Priority (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP08171580A EP2196972A1 (en) | 2008-12-12 | 2008-12-12 | Object tracking system, object infrastructure provided with object tracking system and method for tracking objects |
EP09771423.2A EP2374117B1 (en) | 2008-12-12 | 2009-12-11 | Vehicle tracking system, vehicle infrastructure provided with vehicle tracking system and method for tracking vehicles |
PCT/NL2009/050758 WO2010068106A1 (en) | 2008-12-12 | 2009-12-11 | Vehicle tracking system, vehicle infrastructure provided with vehicle tracking system and method for tracking vehicles |
Publications (2)
Publication Number | Publication Date |
---|---|
EP2374117A1 EP2374117A1 (en) | 2011-10-12 |
EP2374117B1 true EP2374117B1 (en) | 2017-08-30 |
Family
ID=40756284
Family Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP08171580A Withdrawn EP2196972A1 (en) | 2008-12-12 | 2008-12-12 | Object tracking system, object infrastructure provided with object tracking system and method for tracking objects |
EP09771423.2A Active EP2374117B1 (en) | 2008-12-12 | 2009-12-11 | Vehicle tracking system, vehicle infrastructure provided with vehicle tracking system and method for tracking vehicles |
Family Applications Before (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP08171580A Withdrawn EP2196972A1 (en) | 2008-12-12 | 2008-12-12 | Object tracking system, object infrastructure provided with object tracking system and method for tracking objects |
Country Status (2)
Country | Link |
---|---|
EP (2) | EP2196972A1 (en) |
WO (1) | WO2010068106A1 (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7991550B2 (en) * | 2006-02-03 | 2011-08-02 | GM Global Technology Operations LLC | Method and apparatus for on-vehicle calibration and orientation of object-tracking systems |
CN110542885B (en) * | 2019-08-13 | 2021-09-21 | 北京理工大学 | Millimeter wave radar target tracking method in complex traffic environment |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5801943A (en) * | 1993-07-23 | 1998-09-01 | Condition Monitoring Systems | Traffic surveillance and simulation apparatus |
US6697103B1 (en) * | 1998-03-19 | 2004-02-24 | Dennis Sunga Fernandez | Integrated network for monitoring remote objects |
AUPQ684600A0 (en) * | 2000-04-11 | 2000-05-11 | Safehouse International Limited | An object monitoring system |
WO2006068463A1 (en) * | 2004-12-24 | 2006-06-29 | Ultrawaves Design Holding B.V. | Intelligent distributed image processing |
-
2008
- 2008-12-12 EP EP08171580A patent/EP2196972A1/en not_active Withdrawn
-
2009
- 2009-12-11 EP EP09771423.2A patent/EP2374117B1/en active Active
- 2009-12-11 WO PCT/NL2009/050758 patent/WO2010068106A1/en active Application Filing
Non-Patent Citations (1)
Title |
---|
None * |
Also Published As
Publication number | Publication date |
---|---|
EP2196972A1 (en) | 2010-06-16 |
EP2374117A1 (en) | 2011-10-12 |
WO2010068106A1 (en) | 2010-06-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP2370965B1 (en) | Traffic information unit, traffic information system, vehicle management system, vehicle and method of controlling a vehicle | |
Li et al. | Split covariance intersection filter: Theory and its application to vehicle localization | |
Li et al. | Space-time registration of radar and ESM using unscented Kalman filter | |
WO2021058099A1 (en) | Multi-step traffic prediction | |
Finke et al. | Cooperative control via task load balancing for networked uninhabited autonomous vehicles | |
Li et al. | A comparison of detrending models and multi-regime models for traffic flow prediction | |
Yu et al. | Interacting multiple model filter-based distributed target tracking algorithm in underwater wireless sensor networks | |
CN101839973A (en) | Track correlation method by taking topological sequences as characteristics | |
Aljeri et al. | Movement prediction models for vehicular networks: an empirical analysis | |
Gloudemans et al. | Interstate-24 motion: Closing the loop on smart mobility | |
Edelmayer et al. | Cooperative federated filtering approach for enhanced position estimation and sensor fault tolerance in ad-hoc vehicle networks | |
EP2374117B1 (en) | Vehicle tracking system, vehicle infrastructure provided with vehicle tracking system and method for tracking vehicles | |
Li et al. | A short revisit of nonlinear gaussian filters: State-of-the-art and some concerns | |
Ghods et al. | Real-time estimation of turning movement counts at signalized intersections using signal phase information | |
Zhang et al. | Multiple vehicle cooperative localization under random finite set framework | |
Baek et al. | Method for estimating population OD matrix based on probe vehicles | |
Chen et al. | Multitarget multisensor tracking | |
Jayanthi et al. | Traffic time series forecasting on highways-a contemporary survey of models, methods and techniques | |
Farag | Bayesian localization in real-time using probabilistic maps and unscented-Kalman-filters | |
Noack et al. | State estimation considering negative information with switching Kalman and ellipsoidal filtering | |
CN114916059A (en) | WiFi fingerprint sparse map extension method based on interval random logarithm shadow model | |
Goli et al. | Cooperative multi-sensor multi-vehicle localization in vehicular adhoc networks | |
Schubert et al. | Unifying Bayesian networks and IMM filtering for improved multiple model estimation | |
Worrall et al. | A probabilistic method for detecting impending vehicle interactions | |
Nantes et al. | Bayesian inference of traffic volumes based on Bluetooth data |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
17P | Request for examination filed |
Effective date: 20110705 |
|
AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO SE SI SK SM TR |
|
DAX | Request for extension of the european patent (deleted) | ||
RAP1 | Party data changed (applicant data changed or rights of an application transferred) |
Owner name: NEDERLANDSE ORGANISATIE VOOR TOEGEPAST- NATUURWETE |
|
GRAP | Despatch of communication of intention to grant a patent |
Free format text: ORIGINAL CODE: EPIDOSNIGR1 |
|
INTG | Intention to grant announced |
Effective date: 20170322 |
|
RIN1 | Information on inventor provided before grant (corrected) |
Inventor name: SIJS, JORIS Inventor name: THEEUWES, JOHANNES, ADRIANUS, CORNELIS Inventor name: DOODEMAN, GERARDUS, JOHANNES, NICOLAAS Inventor name: PAPP, ZOLTAN Inventor name: NELISSE, MARTIN, WILLEM Inventor name: DRIESSEN, BART |
|
GRAS | Grant fee paid |
Free format text: ORIGINAL CODE: EPIDOSNIGR3 |
|
GRAA | (expected) grant |
Free format text: ORIGINAL CODE: 0009210 |
|
AK | Designated contracting states |
Kind code of ref document: B1 Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO SE SI SK SM TR |
|
REG | Reference to a national code |
Ref country code: GB Ref legal event code: FG4D |
|
REG | Reference to a national code |
Ref country code: CH Ref legal event code: EP |
|
REG | Reference to a national code |
Ref country code: AT Ref legal event code: REF Ref document number: 924283 Country of ref document: AT Kind code of ref document: T Effective date: 20170915 |
|
REG | Reference to a national code |
Ref country code: IE Ref legal event code: FG4D |
|
REG | Reference to a national code |
Ref country code: DE Ref legal event code: R096 Ref document number: 602009048071 Country of ref document: DE |
|
REG | Reference to a national code |
Ref country code: NL Ref legal event code: FP |
|
REG | Reference to a national code |
Ref country code: FR Ref legal event code: PLFP Year of fee payment: 9 |
|
REG | Reference to a national code |
Ref country code: LT Ref legal event code: MG4D |
|
REG | Reference to a national code |
Ref country code: AT Ref legal event code: MK05 Ref document number: 924283 Country of ref document: AT Kind code of ref document: T Effective date: 20170830 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: SE Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 Ref country code: AT Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 Ref country code: FI Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 Ref country code: NO Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20171130 Ref country code: HR Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 Ref country code: LT Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: LV Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 Ref country code: IS Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20171230 Ref country code: BG Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20171130 Ref country code: GR Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20171201 Ref country code: ES Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: PL Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 Ref country code: DK Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 Ref country code: CZ Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 Ref country code: RO Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: IT Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 Ref country code: SK Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 Ref country code: SM Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 Ref country code: EE Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 |
|
REG | Reference to a national code |
Ref country code: DE Ref legal event code: R097 Ref document number: 602009048071 Country of ref document: DE |
|
PLBE | No opposition filed within time limit |
Free format text: ORIGINAL CODE: 0009261 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT |
|
REG | Reference to a national code |
Ref country code: CH Ref legal event code: PL |
|
26N | No opposition filed |
Effective date: 20180531 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: SI Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 |
|
REG | Reference to a national code |
Ref country code: IE Ref legal event code: MM4A |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: MT Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20171211 Ref country code: LU Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20171211 |
|
REG | Reference to a national code |
Ref country code: BE Ref legal event code: MM Effective date: 20171231 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: IE Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20171211 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: BE Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20171231 Ref country code: CH Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20171231 Ref country code: LI Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20171231 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: HU Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT; INVALID AB INITIO Effective date: 20091211 Ref country code: MC Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: CY Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20170830 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: MK Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: TR Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: PT Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20170830 |
|
P01 | Opt-out of the competence of the unified patent court (upc) registered |
Effective date: 20230522 |
|
PGFP | Annual fee paid to national office [announced via postgrant information from national office to epo] |
Ref country code: GB Payment date: 20231220 Year of fee payment: 15 |
|
PGFP | Annual fee paid to national office [announced via postgrant information from national office to epo] |
Ref country code: NL Payment date: 20231220 Year of fee payment: 15 Ref country code: FR Payment date: 20231221 Year of fee payment: 15 Ref country code: DE Payment date: 20231214 Year of fee payment: 15 |