JP4415278B2 - Traffic flow behavior estimation system at intersections - Google Patents

Description

  The present invention relates to traffic flow behavior at intersections as an elemental technology necessary for realizing safe, smooth, or environmentally friendly ground-based signal control and information provision, in-vehicle safe driving and environmental measures. It relates to estimation and prediction technology.

As a technology for estimating the position of the end of the intersection matrix with a measuring device installed on the ground, the occupancy rate (occupation time) at that point obtained by optical beacons, ultrasonic vehicle detectors, and loop vehicle detectors can be used. ) And speed estimation techniques are known, and are used to determine splits in signal control, and to provide traffic length and travel time for navigation purposes in VICS.
Especially on ordinary roads, ultrasonic vehicle detectors are widely used. The installation criteria for ultrasonic vehicle detectors are determined based on the distance from the stop line at the intersection, but the installation of ultrasonic vehicle detectors is very rough with an interval of 100 meters or more. In addition, ultrasonic vehicle detectors are not always installed on local roads according to standards. Therefore, the current situation is that the position and travel time at the end of the matrix with the intersection as the bottleneck cannot be accurately estimated and predicted.

In particular, when the purpose is predictive signal control a few minutes ahead and driving support for safety and environmental measures until the vehicle reaches the downstream intersection, higher prediction accuracy is required, although it is a short time ahead, It is difficult to achieve the objective with the current method.
Although a method of directly recognizing the end of the matrix using an image sensor is conceivable, it is not practical because a large amount of equipment needs to be newly installed on the road side.

On the other hand, with respect to probe data (also referred to as uplink data) collected from the in-vehicle device of the probe vehicle, the travel time used in navigation etc. is estimated by averaging the data collected at a time interval of about 5 minutes. That has been widely studied. For example, Patent Document 1 (travel link determination system and link travel time measurement system) shows a method for measuring travel time, and Patent Document 2 (method and program for determining abnormalities of moving body trajectory information) shows probe data. The method of abnormal processing of the locus is shown. Patent document 3 (link travel time estimation method and apparatus) discloses a method of improving travel time estimation accuracy by using together travel time based on probe data and travel time obtained by a vehicle detector.
JP 2004-101504 A JP 2005-156308 A JP 2004-295165 A

However, using past probe data, etc., the traffic flow behavior such as the stop position in the matrix of the vehicle at the current time or very short time ahead, the stop time in the matrix, and the time required from the stop to the stop line passing through the intersection can be accurately determined. Methods for estimation and prediction have not been studied so far.
The present invention is an intersection that can accurately estimate and predict traffic flow behavior such as a stop position in a matrix of vehicles, a stop time in the matrix, and a required time from the stop to a predetermined point near the intersection of the intersection in real time. The purpose is to provide a traffic flow behavior estimation system.

  The traffic flow behavior estimation system of the present invention acquires probe data acquisition means for acquiring probe data including vehicle position data measured by an in-vehicle device mounted on a vehicle, and signal switching timing information of a traffic light installed at an intersection. A signal switching timing acquisition means, a traffic acquisition means for acquiring traffic data of the road measured by a vehicle detector installed on a road entering the intersection, and a vehicle detector passage time of the vehicle is obtained. A passage time acquisition means, a traffic flow behavior acquisition means for obtaining any one of a stop position in the matrix of the vehicle, a stop time in the matrix, a required time from the stop to a predetermined point passing near the intersection by the probe data, and a vehicle of the vehicle The time from the sensor passage time to the signal switching timing, the traffic volume at the vehicle sensor passage time or the vehicle sensor passage time. Find the correlation between the traffic volume measured once or multiple times retroactively from the time, and the stop position in the vehicle matrix, the stop time in the matrix, and the required time from the stop to the passage of a predetermined point near the intersection. Correlation calculation means and vehicle detector passing time, signal switching timing, traffic volume at the vehicle detector passing time or vehicle detector passing time for any vehicle passing through the road entering the intersection Any one of the stop position in the matrix of the vehicle, the stop time in the matrix, and the required time from the stop to the passage of the predetermined point near the intersection from the traffic volume measured once or multiple times retroactively and the correlation Traffic flow behavior output means for calculating and outputting.

Here, the “predetermined point near the intersection” refers to a stop line at the intersection, for example.
This system statistically analyzes the relationship between the traffic flow behavior of the probe vehicle at the intersection, signal switching timing, and traffic volume, creates the correlation, and measures the created correlation and general vehicle. Based on the traffic volume and the signal switching timing, it is possible to create the estimated information of the traffic flow behavior at the present time of the general vehicle that has passed the vehicle detector or the prediction information of the traffic flow behavior in the near future.

It said correlation computing means divides one signal period of the traffic into a plurality of time zones, fitted to one of the time division time from vehicle detectors passing time of the vehicle to the signal switching timing, the correlation for each time segment It is preferable to obtain a relationship. Thereby, the estimation precision of traffic flow behavior can be improved by setting the memory | storage classification | category of a correlation finely.
Another traffic flow behavior estimating system of the present invention, the probe data acquisition unit and, installed in the traffic signal to the intersection difference points to acquire probe data including the vehicle position data measured by the in-vehicle device mounted in the vehicle Signal switching timing acquisition means for acquiring switching timing information; traffic volume acquisition means for acquiring traffic volume data of the road measured by a vehicle detector installed on the road entering the intersection; and vehicle detection of the vehicle Transit time obtaining means for obtaining the passage time of the vehicle, and traffic flow behavior obtaining means for obtaining any one of the stop position in the vehicle matrix, the stop time in the matrix, and the required time from the stop to the passage of the predetermined point near the intersection by the probe data. And the time from the vehicle sensor passage time to the signal switching timing, the traffic at the vehicle sensor passage time or the vehicle The relationship between the traffic volume measured once or multiple times retroactively from the sensor passage time, and the stop position in the vehicle matrix, the stop time in the matrix, and the time required from the stop to the passage of a predetermined point near the intersection For any vehicle passing through the road entering the intersection, the vehicle detector passage time, the signal switching timing, the traffic volume at the vehicle detector passage time, or the vehicle Based on the traffic volume measured once or multiple times retrospectively from the sensor passage time and the data stored in the storage means, the stop position in the matrix of the vehicle, the stop time in the matrix, the predetermined distance from the stop to the intersection Traffic flow behavior output means for outputting any of the time required to pass the point.

Here, the “predetermined point near the intersection” refers to a stop line at the intersection, for example.
In the aspect of the present invention, the system stores the relationship among the traffic flow behavior of the probe vehicle at the intersection, the signal switching timing, and the traffic volume in the storage means in the form of data. From the data, measured traffic volume and signal switching timing, it is possible to create information on traffic flow behavior at the present time of general vehicles that have passed through the vehicle detector, or prediction information on traffic flow behavior in the near future. .

The storage means divides one signal period of the traffic into a plurality of time zones, fitted to one of the time division time from vehicle detectors passing time of the vehicle to the signal switching timing, the relationship for each time segment Is desirable. Thereby, the estimation accuracy of the traffic flow behavior can be improved by finely setting the storage section of the storage means.
The storage means from the value 0 in traffic, the vehicle detector passing time in traffic or the vehicle detector passed back from the time the maximum value among the plurality of times measured traffic volume traffic to a plurality of The traffic volume is divided into traffic categories, and the traffic volume at the vehicle sensor passage time or the traffic volume measured multiple times retrospectively from the vehicle sensor passage time is selected according to the value of the measured traffic volume . It is desirable that the relationship is stored as data for each traffic volume category, applied to the traffic volume category. Thus, by setting the storage section of the storage means to an appropriate size, the overall storage capacity can be reduced and the access time can be shortened.

  The storage means may change the size of the traffic volume classification according to the number of data in the classification. As a result, the data storage efficiency can be improved as a whole by expanding the section when the number of data in the section is large and decreasing the section when the number of data is small.

The traffic flow behavior prediction information data such as the stop position in the matrix of the vehicle calculated using the present invention, the stop time in the matrix, the required time from the stop to the intersection stop line, the capacity monitoring of the road / intersection, It can be used for daily traffic management such as daily reports and long-term forecasts of traffic volume. It can also be used as one of basic information such as signal control and VICS.
Predictive signal control that uses the calculated prediction information of traffic flow behavior at intersections to determine the necessity of extending the green time of signals, thereby delaying the start of traffic jams and suppressing the occurrence of traffic jams It can be performed. In addition, adaptive signal control can be performed in which a split is determined in proportion to traffic demand at an intersection. Furthermore, traffic flow behavior estimation information can be constantly recorded to evaluate signal control.

  In addition, in the roadside device or traffic control center, the prediction information of the short-term traffic flow behavior is created using the prediction information of the traffic flow behavior calculated in the present invention, and the vehicle about to enter the intersection, Providing prediction information of the traffic flow behavior until the vehicle passes the intersection by road-to-vehicle communication, so that the vehicle travels safely to the intersection (dilemma traveling control, etc.) and environment (idling stop control, etc.) It is possible to provide information and automatic driving control in consideration of

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. First, terms are defined.
“Traffic volume”: Number of vehicles passing per unit time.
“Occupied time”: The total time that the vehicle exists in the detection area of the vehicle detector per unit time.
“Occupancy rate”: Occupation time / unit time “Probe vehicle”: A vehicle equipped with a specific in-vehicle device capable of transmitting data such as vehicle position, speed, time, etc. necessary for carrying out the present invention.

“General Vehicle”: All vehicles including probe vehicles.
“Probe data”: data such as the vehicle position, speed, and time obtained from the in-vehicle device of the probe vehicle. The probe data is data such as vehicle position, speed, time, etc. every predetermined time, every predetermined distance, every predetermined acceleration / deceleration, or every predetermined direction change, and the position of the vehicle when the vehicle is stopped or when the vehicle starts. , Including data such as speed and time.

“Stop queue”: A queue of stopped vehicles waiting for a signal.
“Signal switching timing information”: Information related to the switching time of red, blue, and yellow at the intersection.
“Signal period”: The period from the start of red (blue) of a signal to the start of the next red (blue).
"Traffic flow behavior": Distance from a predetermined point near the intersection to the vehicle stop position in the queue, stoppage time of the vehicle in the stop queue at the intersection, required time from the start of the vehicle to passing the predetermined point near the intersection Say.

FIG. 1 is a map showing an intersection that is a target of traffic flow behavior calculation and a road R (also referred to as a target road) R that flows into the intersection.
An upstream roadside device 5 that performs road-to-vehicle communication with a vehicle is installed at a roadside position upstream of the road R that flows into the intersection. Moreover, the downstream roadside apparatus 5 is installed at the exit of the intersection. These roadside devices 5 can use narrow-band wireless communication such as optical beacons and DSRC, but may use medium-band wireless communication such as a wireless LAN and wide-band wireless communication such as a mobile phone.

The location of the roadside device 5 is preferably on the road R side when using narrow-area wireless communication, but when using mid-range wireless communication, the roadside device 5 can be installed anywhere if road-to-vehicle communication can be performed with the vehicle. Also good.
The roadside device 5 is connected to the traffic control center 3 through a communication line as will be described later with reference to FIG.

The roadside device 5 receives the probe data of the probe vehicle that enters the intersection through the target road R.
Regarding the handling of the probe data, as shown in the embodiment of FIG. 8 later, wide area wireless communication such as a mobile phone mounted on or brought into the probe vehicle from the probe vehicle without passing through the roadside device 5 is performed. It is also conceivable to use the direct transmission to the probe center 3a.

Hereinafter, an example of a stop line at an intersection will be described as “a predetermined point near the intersection”.
The vehicle sensor 6 for measuring the traffic volume is installed in the vicinity of a roadside location 5 in FIG. The distance between the stop line at the intersection and the vehicle detector 6 is Ls. The vehicle detector 6 performs traffic volume measurement at a predetermined cycle. Note that when the roadside device 5 is an optical beacon, the traffic volume can be measured, so the roadside device 5 also serves as the vehicle detector 6.

In addition, a signal control device 4 for controlling the lighting timing of the red, yellow, and blue signals of the traffic light is installed at the intersection.
FIG. 2 is a graph showing the running behavior of the vehicle at the intersection, where the horizontal axis represents time and the vertical axis represents the distance from the stop line of the intersection. Let red signal start time be time tr and green signal start time be tg.

  The vehicle identification number data and probe data of the probe vehicle that enters the intersection through the target road R are received by the roadside device 5 and accumulated in the roadside device 5. This information may be accumulated in the traffic control center 3 through the roadside device 5. When this data is analyzed, the time when the probe vehicle passes through each point of the inflow road R, and the position and speed at each time are obtained. Thereby, the distance and speed from the stop line of the intersection for every time of the probe vehicle can be grasped, and the traveling locus U of the probe vehicle as shown in FIG. 2 is obtained.

Also, based on the signal switching timing information obtained from the signal control device 4, information on the start time tr of the red signal (which may include a yellow signal; the same shall apply hereinafter) and the start time tg of the blue signal are also obtained. These times are also entered in the graph.
A triangular portion represented by a vertical line in FIG. 2 represents a stop queue of the vehicle. The upper side of the triangle represents the stop position of the vehicle that has entered the stop queue. In the example of FIG. 2, the vehicle stops at the end of the red light stop queue. The diagonally lower side of the triangle represents a start position of a vehicle that turns green and starts from an intersection. When it turns green, the vehicle starts from the top of the stop queue. Since the stop position and the start position each extend upstream with time, each has a propagation velocity. These are called “stop wave” and “start wave”.

The point at which the stop wave and the start wave cross represents a boundary point where the stop queue gradually decreases while the stopped vehicle travels in the stop queue. Vehicles that enter the intersection after this time travel in the procession of moving vehicles.
It is assumed that the time when the probe vehicle passes the vehicle detector 6 is tp, the time when the probe vehicle enters the stop queue and stops is ts, and the time when the probe vehicle starts from the head of the stop queue is tm. Let Lp be the distance from the stopping probe vehicle to the stop line at the intersection. In FIG. 2, the probe vehicle arrives at the end of the stop queue at time ts at a distance Lp from the intersection stop line, stops, and then the signal turns blue at time tg, toward the intersection at time tm. It can be seen that it started moving and passed the stop line at the intersection.

In this way, depending on the degree of traffic congestion, traffic volume, and signal switching timing information, the traveling vehicle may stop at the end of the stop queue waiting for traffic lights, or may add to the end of a moving queue with a green light. Or, if the vehicle passes smoothly through the intersection without adding to the stop queue length, various traveling tracks can be obtained.
Hereinafter, an example of the distance L from the stop line at the intersection to the vehicle stopped in the matrix will be described as an example of “traffic flow behavior”. It is assumed that there is no clogging downstream after passing through the intersection and there is a free flow.

Here, the stop position L in the matrix measured from the stop line of the intersection is the time T from tp to tr and the traffic volume Q measured by the vehicle sensor 6 at the time tp as shown in the following (formula 1). I think that it becomes a function.
L = f (T, Q) (1 formula)
The traffic volume Q requires the traffic volume of this cycle and a cycle measured before this cycle, based on the cycle before the nearest time (latest) from the time tp when the probe vehicle passes the vehicle detector 6. You may accumulate and use for hours. Further, based on the period of time including the time tp when the probe vehicle passes the vehicle sensor 6, the traffic volume of this period and the period measured before this period may be accumulated for the necessary time.

The time from the time tp when the probe vehicle passes through the vehicle detector 6 to the red start time tr of the signal is Tp (Tp = tr−tp), and the probe vehicle stops in the matrix measured from the stop line at the intersection. Let the stop position be Lp. The traffic volume measured at time tp when the probe vehicle passes the vehicle detector 6 is defined as Qp. Applying these to (Equation 1)
Lp = f (Tp, Qp) (2 formulas)
It becomes.

For a plurality of probe vehicles, the data of Tp, Qp, and Lp are totaled. Then, based on the tabulated data, an attempt is made to obtain the function form f of the (Expression 1).
For this purpose, first, one period C of the signal is set to (T0, T1), ..., (Ti-1, Ti), ..., (Tm-1, Tm) (1≤i≤m, T0 = 0). , Tm = C). Then, it is determined which of these divided data ranges the time Tp from the time tp when the probe vehicle passes the vehicle sensor 6 to the red start time tr of the signal. However, it is assumed that 0 ≦ Tp ≦ C and C is one signal period.

In FIG. 3, one period C of the signal is divided into m by the data range (T0, T1),..., (Ti-1, Ti),. It is a figure which shows a mode that it is in the data range (T2, T3).
The past data Qp and Lp are statistically analyzed for each data range (Ti-1, Ti) of one period C of the signal so that L is uniquely determined for any appearance value of Q. The following processing is performed.
<Method 1> (Correlation analysis)
In each data range (Ti-1, Ti) of T, it is considered that L and Q have the following correlation.
<Method 1-1> (Single correlation analysis)
In each data range (Ti-1, Ti), when L and Q are related to T0≤Tp <T1, L = a1Q + b1 (Equation 3-1)
When T1≤Tp <T2 L = a2Q + b2 (Equation 3-2)
...
When Tm-1≤Tp <Tm L = amQ + bm (3-m formula)
Is assumed. a1 to am and b1 to bm are called correlation parameters.

For example, it is assumed that the time Tp from the time when a certain probe vehicle passes the vehicle sensor 6 to the red start time tr of the signal is in the range of T0 ≦ Tp <T1. The stop position in the matrix measured from the stop line of the intersection is Lp, and the traffic volume measured at time tp is Qp. A lot of such data is collected, and around the regression line of (Equation 3-1) Correlation parameters a1 and b1 and unbiased variance σ1 ² that minimize the square error of data are calculated.

It is assumed that the time Tp from the time when another probe vehicle passes the vehicle sensor 6 to the red start time tr of the signal is in the range of T1 ≦ Tp <T2. The stop position in the matrix measured from the stop line at the intersection is Lp, and the traffic volume measured at time tp is Qp. A lot of such data is collected, and around the regression line of (Equation 3-2) Correlation parameters a2 and b2 and unbiased variance σ2 ² that minimize the square error of data are calculated.

In this manner, the correlation parameters a1, b1, unbiased variance .sigma.1 ^2, the correlation parameter a2, b2, unbiased variance .sigma. @ 2 ^2, · · · correlation parameters am, bm, an unbiased variance .sigma.m ² obtains. That is, the correlation parameter and unbiased variance are calculated independently over the entire data range of Tp.
<Method 1-2> (Multiple correlation analysis)
Next, the multiple correlation analysis process in the case of using the traffic volume measured a plurality of times by tracing back the time tp with the vehicle detector 6 will be described.

As shown in FIG. 4, the traffic volume Q1, -----, tp-tj to tp-tj measured from tp-t1 to tp with reference to the time tp when the probe vehicle passes the vehicle detector 6. measured by -1 the traffic Qj, --------, so on tp-t _n from tp-t _n-1 until the measured traffic volume Qn, a plurality n before time tp The traffic volume measured (n ≧ 2) times is used. Here, tj (1.ltoreq.j.ltoreq.n, t0 = 0) is a predetermined time, and is set to, for example, the signal period C and a multiple thereof.

L = a (1,1) Q1 + a (1,2) Q2 + when the expression T0 ≦ Tp <T1 including the correlation parameters aj, b in each data range (Ti-1, Ti) of one period C of the signal --- a (1, j) Qj + --- + a (1, n) Qn + b1 (Formula 4-1)
When T1≤Tp <T2 L = a (2,1) Q1 + a (2,2) Q2 ++-a (2, j) Qj + --- + a (2, n) Qn + b2 (Formula 4-2)
...
When Tm-1≤Tp <Tm L = a (m, 1) Q1 + a (m, 2) Q2 ++-a (m, j) Qj + --- + a (m, n) Qn + bm (4-m formula)
Is assumed. However, 1 ≦ j ≦ n.

For example, the time Tp from the time when a certain probe vehicle passes the vehicle sensor 6 to the red start time tr of the signal falls within the data range of T0 ≦ Tp <T1, and is within the matrix measured from the stop line at the intersection at that time. When the stop position is Lp and each traffic volume measured a plurality of times by the vehicle detector 6 before that time is Q1, Q2,. correlation parameter a the square error of around data regression line to minimize the) (1,1) ~a (1, n), calculates the b1 and unbiased variance .sigma.1 ^2.

The time Tp from the time when the other probe vehicle passes the vehicle sensor 6 to the red start time tr of the signal falls within the data range of Tm-1 ≦ Tp <Tm, and is within the matrix measured from the stop line at the intersection at that time The stop position is Lp, and the traffic volume measured by the vehicle detector 6 several times by that time is Q1, ..., Qn, and many such data groups are collected, and the regression line of (4-2) Correlation parameters a (m, 1) to a (m, n), bm and unbiased variance σm ² that minimize the square error of the data around are calculated.

In this manner, the correlation parameters a (1,1) to a (1, n), b1 and unbiased variances σ1 ^{2 to a} (m, 1) to a (m, n), bm and unbiased variance σm ² are calculated.
The unbiased variance σ ² for each data range of Tp is calculated as follows, for example. Let the vector of the occurrence data Li of L, which is the objective variable, be y as in the following equation. Here, (*) ^T represents transposition of (*).
y = (L1, L2, ---, Ln) ^T
In addition, the regression estimated value of each occurrence data by the coefficients a (i, 1), a (i, 2),-(), a (i, n) is Yi, and the vector is Y.
Y = (Y1, Y2, ---, Yn) ^T
Yi = a (i, 1) Q1, i + a (i, 2) Q2, i + --- a (i, j) Qj, i + --- + a (i, n) Qn, i + bi
At this time, the fitting error (variation of residual from the regression) EV in the regression analysis can be calculated by the following equation.
EV = Σ (Li−Yi) ² = (y−Y) ^T (y−Y)
Therefore, EV may be considered as the variance of the multiple correlation analysis, and its square root as the standard deviation.

In this <method 1>, the correlation parameter is stored for each data range (Ti-1, Ti), so that the storage memory of the storage memory is compared with the case of storing data as in <method 2> and <method 3> described later. Capacity is reduced.
Depending on how busy the traffic is, change the number of divisions T, the number of times of adoption of traffic volume Q (how many times the measurement value is used) n, and the adoption measurement interval tj of traffic volume Q. Also good. For example, as the traffic is congested, the division number m is decreased, the employed measurement number n is increased, and the employed measurement interval tj is increased. On the contrary, if the traffic is free, the division number m may be increased, the adopted measurement number n may be reduced, and the adopted measurement interval tj may be shortened. The degree of traffic congestion can be evaluated by, for example, determining a threshold value and comparing the magnitudes of Lp and Qp with the threshold value.

Since the correlation parameters aj and bj do not change greatly in a short time, they may be averaged or smoothed in consideration of data calculated in the past.
In the above method, an example of formulating a linear (primary) correlation is shown, but a nonlinear formulation such as a relationship consisting of two straight lines or a quadratic equation may be used. In the above description, the correlation is calculated regardless of the time. However, the correlation may be calculated in units of time.
<Method 2> (Pattern classification)
There is a method of classifying L for each pattern as follows when the accuracy of formulation of Lp and Q is low (for example, when the linear relationship is not always guaranteed) or when there is a margin in the storage memory.

In this method, as in <Method 1-2>, the traffic volume Qp measured by the vehicle detector 6 a plurality of times after the time tp is used. Traffic Q1 from tp-t1 to tp,..., Tp-tj to tp-tj-1, with reference to the time tp when the probe vehicle passes the vehicle detector 6. The traffic volume Qn (n ≧ 2) from −t _n to tp−t _{n−1 is} assumed.
By such multiple measurements, an (n + 1) -dimensional space consisting of time T and traffic volumes Q1, Q2,..., Qn is formed.

In the present <Method 2>, as described above, one period C of the signal is divided into m, and the time Tp from the time tp when the probe vehicle passes the vehicle sensor 6 to the red start time tr of the signal is divided into these divisions. Into one of the specified data ranges.
Furthermore, in this <Method 2>, each traffic volume Qj is divided into data ranges (Qj, 0, Qj, 1), --Q-,-(Qj, k-1, Qj, k),... (Qj, r- 1. Divide into r, Qj, r). However, 1 ≦ j ≦ n, 1 ≦ k ≦ r, Qj, 0 = 0, Qj, r = maxQj, and maxQj is the traffic volume Qj measured each time each probe vehicle passes the vehicle detector. Of these, the maximum value.

In this way, the (n + 1) dimensional space can be divided by the data range.
Table 1 shows an example of a storage format in which stop positions in the matrix of the probe vehicle are stored by dividing each data range. Since Qj measured n = 3 times from the time tp is used, a four-dimensional space composed of T, Q1, Q2, and Q3 is formed. The time T is divided into five data ranges, and the data division number r of each Qj (1 ≦ j ≦ 3) is set to 4.

  For each probe vehicle, the stop position Lp in the matrix measured from the intersection stop line at which the probe vehicle stopped, and the time T from the time tp when the probe vehicle passed the vehicle detector 6 to the red start time tr of the signal. And the traffic volume Qp measured three times retrospectively from the time tp, the data range (Ti-1, Ti), (Q1, k-1, Q1, k), (Q2, k-1, Q2, k), (Q3, k-1, Q3, k).

  For example, the time Tp from the time when a certain probe vehicle passes the vehicle detector 6 to the red start time tr of the signal enters the second (T1, T2) data range, and the vehicle detector 6 before the passing time. Assume that each traffic volume measured multiple times is Q1, Q2, and Q3. Assume that Q1 enters the second data range, Q2 enters the first data range, and Q3 enters the second data range. In Table 1, these data ranges “2212” are circled.

For the other probe vehicles, the stop positions Lp in the matrix measured from the stop line at the intersection where the probe vehicle stopped are stored in each data range of the four-dimensional space in the same manner as described above.
When data of a predetermined number of probe vehicles are collected, L is extracted within each data range, and the average value and variance are calculated.
Table 2 shows the number of data of the stop position L in the matrix of the probe vehicle extracted within each data range, the average value of L, and the variance of L. Here, “the presence / absence of division” and “the number of divisions” will be described in <Method 3>.

In this method, the function form is not limited as in <Method 1>, but a large storage capacity is required for the memory. In the above example, an example is shown in which the traffic volume Qp is measured a plurality of times. In this case, a two-dimensional space of T and Q is obtained.
<Method 3> (Adaptive pattern classification)
In this method, in the <method 2>, the division number m of the data range (Ti-1, Ti) of T and Q and the division number r of (Qj, k-1, Qj, k) are not fixed in advance. In the data processing process, the data range is changed from the coarse range to the fine range, or from the fine range to the coarse range according to the variation in the number of probe data for each data range and the numerical value of L.

For example, when the pattern of the data range of T and Qj in Table 1 is “2212”, if the number of L data in this data range and the variance of the numerical values of L are equal to or greater than a predetermined threshold, the data is limited to this pattern. Subdivide the range into a, b, c,. If the average values of L are substantially the same, the data ranges are integrated.
When subdivision is performed, “division presence / absence” is set to 1, and the number of divisions is recorded in [Table 2]. When the subdivision is not performed, “division presence / absence” is zero.

  Table 3 shows an example of the re-divided / integrated table storage pattern.

  In Table 3, an example in which the second (T1, T2) data range of time T is divided into three, the second data range of traffic volume Q1 is divided into two, and the first data range of traffic volume Q2 is divided into four. Is shown. Further, the third and fourth data ranges of the traffic volume Q3 are integrated. Therefore, the division number is “3, 2, 4, 1”, and “3, 2, 4, 1” is recorded in the pattern “2212” column of [Table 2].

  The number of divisions of the data range may be set with variable length for each data based on the numerical distribution of T, Qj data, etc. (Table 3 shows an example of setting with variable length) or for each data It may be divided into fixed lengths. In this way, by performing re-division / integration once to multiple times, it is possible to perform storage with an optimal number of divisions. Therefore, in this method, the storage memory can be greatly reduced as compared with <Method 2>.

  As can be seen from the above analysis, the time T = tr−tp from the time when the vehicle passes the vehicle detector 6 to the red start time tr of the signal, the traffic volume data measured at the time when the vehicle passes the vehicle detector 6. From Q (or traffic data Q1, Q2, -----, Qn measured n times retroactively from that time) and correlation parameter data or table data, the stop position in the general vehicle matrix and its The variance can be estimated. It is also possible to predict stop positions and their variances in the near future matrix.

  When the slope of the starting wave shown in FIG. 1 is a known value (usually, the slope of the starting wave can be regarded as constant), the stop time in the matrix (tm− ts) can also be estimated. In addition, when the start speed in the matrix is known (usually, the start speed in the matrix can be regarded as constant), using this start speed, the required time from the start to the passage of the stop line at the intersection (tu -Tm) can also be estimated.

As described above, if one of the stop position in the matrix, the stop time in the matrix, and the required time from the start to the passage of the stop line at the intersection are obtained, the other can be easily obtained.
Therefore, in the embodiments of the invention described above, the calculation method of the stop position in the matrix has been described. However, as an indicator of traffic flow behavior, not only the stop position in the matrix but also the stop time in the matrix, the intersection of the target road The time required to pass the stop line may be calculated.

  The predicted stop position in the matrix, the stop time in the matrix, or the predicted value and variance of the required time from the start to the passage of the stop line at the intersection can be received using the roadside device (the vehicle can be received by the in-vehicle device). Information). Therefore, if the vehicle waits for a signal at the intersection, it can predict how long it should wait from the stop line of the intersection, how long it must wait in the time queue, and when it can pass through the intersection it can.

FIG. 5 is a flowchart showing the entire procedure of the traffic flow behavior calculation processing of the present invention described so far.
All or part of the processing described below is realized by a system processing device executing a program recorded on a predetermined medium such as a CD-ROM or a hard disk. In addition, the installation place of the processing apparatus of the system is arbitrary, and as will be described later, it may be installed in the roadside apparatus 5, may be installed in the traffic control center 3, or may be installed in the probe center 3a. There is also.

First, the processing device acquires probe data of the probe vehicle, signal switching timing information, and information on the traffic volume measured by the vehicle detector 6 (step S1).
Next, using <Method 1>, <Method 2> or <Method 3> described above, the distance from the stop line at the intersection to the stop position in the queue, the stop time in the stop queue at the intersection, or after starting A statistical analysis is performed on the relationship between traffic flow behavior, which is the time required from the stop to the stop line passing through the intersection, signal switching timing, and traffic volume (step S2). Then, correlation parameters are obtained as a result of analysis, or table data is created (step S3).

Next, referring to correlation parameters or table data, signal switching timing, and traffic volume (step S4), the following processing is performed.
First, based on the traffic volume measured by the vehicle detector, the traffic flow behavior is estimated using the correlation parameter or table data, and prediction information is created (step S5).
Next, it is determined whether or not the information is used for providing information such as navigation (step S6). If it is used, it is used for calculating the length of traffic jam, the degree of traffic jam, and the travel time at the intersection (step S7).

Next, it is determined whether or not it is used for effective signal control (step S8), and if it is used, the presence or absence of blue extension in predictive signal control, determination of traffic demand in adaptive signal control, signal control (Step S9).
Next, it is determined whether or not to use for providing information to the vehicle and vehicle driving control (step S10), and if used, the traffic flow behavior prediction information until reaching the intersection is used for road-to-vehicle communication. To provide information to the vehicle (step S11). In the vehicle, the traffic flow behavior is used to provide the driver with information that is safe and environmentally friendly, or the vehicle is automatically driven (step S12).
-System configuration example-
Hereinafter, a configuration example of the traffic flow behavior estimation system of the present invention will be described.

First, probe data and the like are transmitted from the probe vehicle C to the roadside device 5, signal switching timing information is transmitted from the signal control device 4 to the roadside device 5, and the roadside device 5 calculates traffic flow behavior estimation and prediction information at the intersection. The system configuration when doing this will be described.
FIG. 6 is a block diagram showing the overall configuration of the traffic flow behavior estimation system of the present invention.
The traffic flow behavior estimation system includes an in-vehicle device 2 of a probe vehicle C, a traffic control center 3, a signal control device 4, and a roadside device 5.

The probe vehicle C includes a communication device for performing road-to-vehicle communication with the roadside device 5, a position detection device for collecting probe data, an information providing unit for providing information in consideration of safety and the environment to the driver, A vehicle-mounted device 2 including a vehicle travel control unit that performs vehicle travel control is mounted.
The said position detection apparatus is equipped with the GPS receiver etc., and can detect the position of the own vehicle for every time. The acquired position data and the like are transmitted from the communication device as probe data. The transmitted data is received by the roadside device 5.

The vehicle detector 6 measures the traffic volume at the place where it is installed. When the roadside device 5 is an optical beacon, the vehicle detector 6 may be included in the roadside device 5.
The roadside device 5 is provided with a communication device for communicating with the in-vehicle device 2 and a processing device. As shown in FIG. 6, the processing apparatus is installed in the roadside apparatus 5, but this is not always necessary, and the processing apparatus may be installed in a place different from the roadside apparatus 5. When the processing device acquires probe data, acquires signal switching timing information from the signal control device 4, and acquires traffic information from the vehicle detector 6, information for predicting traffic flow behavior (correlation parameters or table data). ) Is calculated. Then, traffic flow behavior is estimated and predicted.

The signal control device 4 performs prediction signal control and adaptive signal control using the prediction result of traffic flow behavior.
The traffic control center 3 communicates with the roadside device 5 to obtain traffic flow behavior estimation and prediction data from the roadside device 5. The traffic control center 3 performs statistical analysis and uses it for daily traffic management, or creates information to be provided for navigation or the like.

Note that reception of probe data and transmission of traffic flow behavior prediction information may be performed by different roadside devices 5.
Next, the probe data is transmitted from the probe vehicle C to the traffic control center 3 via the roadside device 5, the traffic volume information is transmitted from the vehicle detector 6 to the traffic control center 3, and the signal switching timing information is signal controlled. A system configuration in the case where a correlation parameter or table data is generated from the device 4 to the traffic control center 3 and created in the traffic control center 3 will be described.

FIG. 7 is a block diagram showing the overall configuration of this traffic flow behavior estimation system.
The difference between the configuration of FIG. 7 and FIG. 6 will be described. Signal switching timing information is directly transmitted from the signal control device 4 to the traffic control center 3, and traffic information is directly transmitted from the vehicle detector 6 to the traffic control center 3. The probe data is transmitted from the roadside device 5 to the traffic control center 3.

The processing device of the traffic control center 3 receives the traffic flow behavior prediction information (correlation parameter or table data), estimates the traffic flow behavior created by the roadside device, and receives the prediction information and provides it to navigation etc. Create information to be used.
Next, the probe data is transmitted from the probe vehicle C to the traffic control center 3 via the probe center 3a, the signal switching timing information is transmitted directly from the signal control device 4 to the traffic control center 3, and the traffic volume information is transmitted to the vehicle. A system configuration in the case where the correlation parameter or the table data is generated directly from the sensor 6 to the traffic control center 3 will be described.

FIG. 8 is a block diagram showing the overall configuration of this traffic flow behavior estimation system.
Explaining the difference between the configuration of FIG. 8 and FIG. 7, there is a probe center 3a that accumulates probe data by performing wide-area communication between the probe vehicle C and a mobile phone or the like. Probe data is supplied from the probe center 3 a to the traffic control center 3.
The processing device of the traffic control center 3 uses this probe data to calculate information (correlation parameters or table data) for predicting traffic flow behavior, and to calculate the traffic flow behavior created by the roadside device, as described above. Receives estimation and prediction information, and creates information to be provided to navigation and the like.

  Although the embodiments of the present invention have been described above, the embodiments of the present invention are not limited to the above-described embodiments, and various modifications can be made within the scope of the present invention.

It is a map depicting the intersection and the target road R entering the intersection. It is a graph showing the time change of a stop queue, and the behavior of the vehicle which approachs an intersection. It is the figure which plotted time Tp from the time tp when the probe vehicle passed the vehicle detector 6 to the red start time tr of the signal within one period C of the signal divided into m. It is a graph which shows the traffic volume Q measured in multiple times with the vehicle sensor. It is a flowchart which shows the whole procedure of the traffic flow behavior calculation process of this invention. 1 is a block diagram showing an overall configuration of a traffic flow behavior estimation system according to an embodiment of the present invention. It is a block diagram which shows the whole structure of the traffic flow behavior estimation system which concerns on other embodiment of this invention. It is a block diagram which shows the whole structure of the traffic flow behavior estimation system which concerns on other embodiment of this invention.

Explanation of symbols

2 On-vehicle device 3 Traffic control center 4 Signal control device 5 Roadside device 6 Vehicle detector

Claims (7)

Probe data acquisition means for acquiring probe data including vehicle position data measured by an in-vehicle device mounted on the vehicle;
Signal switching timing acquisition means for acquiring signal switching timing information of a traffic light installed at an intersection;
A traffic volume acquisition means for acquiring traffic volume data of the road measured by a vehicle detector installed on the road entering the intersection;
Passage time obtaining means for obtaining a vehicle detector passage time of the vehicle;
Traffic flow behavior acquisition means for obtaining any of the stop position in the vehicle matrix, the stop time in the matrix, and the required time from the stop to the passage of the predetermined point near the intersection by the probe data;
The time from the vehicle sensor passage time to the signal switching timing of the vehicle, the traffic volume at the vehicle sensor passage time or the traffic volume measured one or more times retroactively from the vehicle sensor passage time, and the vehicle A correlation calculation means for obtaining a correlation with any one of a stop position in the matrix, a stop time in the matrix, and a required time from the stop to a predetermined point near the intersection;
Once for any vehicle passing through the road entering the intersection, the vehicle detector passing time, the signal switching timing, the traffic volume at the vehicle detector passing time or the vehicle detector passing time once Alternatively, calculate and output either the stop position in the matrix of the vehicle, the stop time in the matrix, or the required time from the stop to the passage of a predetermined point near the intersection from the traffic volume measured multiple times and the correlation. A traffic flow behavior estimation system at an intersection having traffic flow behavior output means. It said correlation computing means divides one signal period of the traffic into a plurality of time zones, fitted to one of the time division time from vehicle detectors passing time of the vehicle to the signal switching timing, the correlation for each time segment The traffic flow behavior estimation system at an intersection according to claim 1, wherein the relationship is obtained. Probe data acquisition means for acquiring probe data including vehicle position data measured by an in-vehicle device mounted on the vehicle;
A signal switching timing acquisition means for acquiring a signal switching timing information of the installed traffic signal to the intersection difference point,
A traffic volume acquisition means for acquiring traffic volume data of the road measured by a vehicle detector installed on the road entering the intersection;
Passage time obtaining means for obtaining a vehicle detector passage time of the vehicle;
Traffic flow behavior acquisition means for obtaining any of the stop position in the vehicle matrix, the stop time in the matrix, and the required time from the stop to the passage of the predetermined point near the intersection by the probe data;
The time from the vehicle sensor passage time to the signal switching timing of the vehicle, the traffic volume at the vehicle sensor passage time or the traffic volume measured one or more times retroactively from the vehicle sensor passage time, and the vehicle Storage means for storing, as data, a relationship between any of the stop position in the matrix, the stop time in the matrix, and the required time from the stop to the passage of a predetermined point near the intersection;
Once for any vehicle passing through the road entering the intersection, the vehicle detector passing time, the signal switching timing, the traffic volume at the vehicle detector passing time or the vehicle detector passing time once Alternatively, any one of the stop position in the matrix of the vehicle, the stop time in the matrix, and the required time from the stop to the passage of the predetermined point near the intersection from the traffic volume measured multiple times and the data stored in the storage means A traffic flow behavior estimation system at an intersection having traffic flow behavior output means for outputting The storage means divides one signal period of the traffic into a plurality of time zones, fitted to one of the time division time from vehicle detectors passing time of the vehicle to the signal switching timing, the relationship for each time segment The traffic flow behavior estimation system at an intersection according to claim 3, which is obtained. The storage means from the value 0 in traffic, the vehicle detector passing time in traffic or the vehicle detector passed back from the time the maximum value among the plurality of times measured traffic volume traffic to a plurality of The traffic volume is divided into traffic categories, and the traffic volume at the vehicle sensor passage time or the traffic volume measured multiple times retrospectively from the vehicle sensor passage time is selected according to the value of the measured traffic volume . traffic behavior estimating system in the traffic segment to fit, traffic classification is for storing as data the relationship for each claim 3 or claim 4 wherein the intersection.   5. The traffic flow behavior estimation system at an intersection according to claim 4, wherein the storage means makes the size of the time section variable according to the number of data in the section.   6. The traffic flow behavior estimation system at an intersection according to claim 5, wherein the storage means makes the size of the traffic volume section variable according to the number of data in the section.

Images