JPH02162499A - Traffic flow sensor with learning function - Google Patents

Abstract

PURPOSE:To immediately detect whether a vehicle going into a signal intersection is dangerous or not in the signal intersection by collecting passing/stop data and data of a position and speed at a suitable updating tie interval, generating a new decision data table by self-learning in a dilemma state varied together with an environmental condition, etc., and updating the data table of a collating part. CONSTITUTION:A position/speed arithmetic part receives a signal of a yellow signal start time point from a control machine of a signal installed in an intersection, calculates the existence distribution and the speed distribution of vehicles existing in an intersection inflow part at the yellow signal start time point, based on the output of a vehicle detecting sensor, and provides an output. Its output is sent to a collating part in order to collate it with a dilemma state decision data table, and by this collation with the table, whether the vehicle is in a dilemma state or not is decided in a read time. In such a manner, the dilemma state varied together with a traffic condition can be recognized in a traffic flow sensor, and in a signal intersection, whether the vehicle going into the intersection is dangerous or not can be detected in a real time.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a traffic flow sensor having a learning function for use in a traffic control system register.

[Prior Art] Traffic accidents have increased in 2015, and more than 60% of the accidents occur near intersections, and countermeasures are strongly desired. The cause of this accident is said to be the dilemma of a vehicle entering an intersection at a yellow light.

The present inventors previously found that they were in such a dilemma +
, developed a method for detecting
It is proposed as No. 9398. This previously proposed method is
The position and speed of a vehicle passing through a signalized intersection inlet are set to the Δl side at the same time as the yellow light of the traffic light turns on or after that Iij, and these measured values are compared with a dilemma state daple created in advance to determine the dilemma state. This detects the presence of heavy objects.

The present inventors also proposed a characteristic measuring device for vehicle dilemma phenomena in Japanese Patent Application No. 63-72840.
C. This device sets measurement points at appropriate intervals within the signal or difference point inlet and intersection, detects passage through JJ at a total of 31!1 points as an electrical signal, and calculates the generation point of the signal and image signal. The system measures the position and speed of a vehicle traveling through the intersection at the start of the color signal, as well as whether the vehicle has passed through the intersection or stopped, and outputs it as necessary data. It is possible to measure the characteristics of dilemma situations that change with environmental and traffic conditions.

On the other hand, real-time control at intersections for traffic control, which has been in practical use for a long time, involves a control center grasping the traffic situation from the output of sensors placed at various locations, and controlling the necessary traffic lights etc. to ensure smooth traffic flow. I'm trying to make this happen.

In the existence of such a control system for traffic control, it is possible to automatically detect whether there is a vehicle in a dilemma state using the previously proposed means described above, and use this to improve traffic safety at signalized intersections, etc. When attempting to reflect this in traffic control, the characteristics of the dilemma described in F vary significantly depending on the environmental conditions and conditions specific to each intersection, so it is necessary to understand the wide-scale traffic situation. Similarly to sensor outputs, it is not an advantageous method to send the measurement results related to the dilemma state to the control center and operate traffic lights etc. as part of the control at the center.

[Invention or problem to be solved] The uninvented technical problem is to make it possible for a traffic flow sensor to grasp the dilemma state that changes with environmental conditions and traffic conditions through self-learning, thereby controlling the results of C5 measurement each time. An object of the present invention is to obtain a traffic flow sensor that is capable of detecting in real time whether a vehicle entering a signalized intersection is dangerous or not without transmitting data to a center.

[Means for Solving the Problems] In order to solve the problems listed in L, the traffic flow sensor of the present invention includes the following:
1. On the road surface of the inlet of the signalized intersection. + A vehicle detection sensor capable of measuring the position and speed of a vehicle passing through the object, and at the same time detecting whether the vehicle has passed through an intersection or a vehicle, and the signal at the start of a yellow light. received,
From the output of the vehicle detection sensor listed in L, the position and speed of the vehicle at the entrance of the intersection are calculated and output at the start of the yellow light. A verification unit that compares the output of the speed calculation unit with the data table; and a verification unit that determines whether or not the vehicle is in a dilemma state based on the output of the verification unit, and generates a signal for safe control of vehicle traffic at the intersection. Danger of outputting
4iI! detects whether the detected vehicle has passed through the intersection or not from the output of the safety identification unit and the heavy vehicle detection sensor as a passing/stop data! The S/Stop+F calculation unit collects the above passing/stop data and the position and speed data obtained in the position/speed rit calculation unit at appropriate update time intervals to create a dilemma state that changes with environmental conditions and traffic conditions. The system also includes a learning section that creates a new dilemma state determination data table through self-learning, and updates the data table of the matching section.

[Operation] When the traffic light changes to yellow, the position and speed calculation unit measures the position and speed of the vehicle at the entrance of the intersection from the output of the vehicle detection sensor, and these measured values are sent to the verification unit.
C. It is compared with pre-created dilemma state determination data, and it is detected in real time whether there is a vehicle in a dilemma state. In the Danger/Safety Judgment Department.

Based on the output of the verification section, it is determined whether the vehicle is in a dilemma state or not, and an essential signal is output for safe control of vehicle traffic at the intersection.

- In the passing/stopping calculation section, whether or not the detected vehicle has passed through the intersection is detected as passing/stopping data from the output of the vehicle detection sensor.

In the learning section, this passing/stopping data and the position and speed data obtained in the position n/velocity calculation section are collected at appropriate update time intervals, and from the data during that time, dilemmas that change with environmental conditions and traffic conditions are collected. A new data table for dilemma state determination is created by self-learning of the state, and the data table of the matching section is updated.

As a result, for vehicles at the entrance of the intersection in -10,000,
While determining whether or not they are in a dilemma state and performing necessary safety control, on the other hand, a new dilemma state determination data table is created through self-learning, and the data table of the collation unit is compared to environmental conditions and traffic conditions. updated according to changes in

Therefore, it is possible to detect in real time whether a vehicle entering a signalized intersection is dangerous or not, without having to send the measurement results to a control center each time. In response to changes in road traffic and traffic conditions, the vehicle will have intelligent functions that will enable it to determine whether a vehicle is entering an intersection, is in a dangerous situation, or is a vehicle.

[Example] FIG. 1 shows the configuration of a traffic flow sensor equipped with one learning machine based on the present invention.

In this traffic flow sensor, as a vehicle detection sensor,
ITV, (:(:D) An image sensor consisting of a group of photoelectric conversion elements, etc., or various sensors using a loop antenna, an ultrasonic sensor, etc. can be used, and these can detect the position and speed of the vehicle at the start of the yellow light. In order to measure
The road surface of the signalized intersection Majinbu will be the subject of measurement, and a measurement area will be set up along that road surface. Furthermore, in order to detect whether the vehicle has passed through the intersection or stopped before the intersection, at least one measurement area is set within the intersection on the extension of the lane to be measured.

The position/velocity calculation unit receives a signal at the start of the yellow light from a traffic light controller installed at the intersection, and calculates the vehicle existing at the inlet of the intersection at the start of the yellow signal based on the output of the vehicle detection sensor. It calculates and outputs the presence distribution and velocity distribution of Based on the comparison, it is determined whether the vehicle is in a dilemma state or not in real time. ! As mentioned above, the signal at the start of 11 points is not limited to the signal from the traffic light controller installed at the intersection, but also the signal change by various other signals such as ITV etc. It is possible to detect this and output it as a signal at the start of the yellow light, or to use a yellow occurrence detector using a photoelectric element.

The above-mentioned dilemma state is an area where vehicles that pass within the yellow light time and full red time at the inflow part of a signalized intersection and vehicles that stop at the stop low line coexist.
No matter which option you choose to stop, you are experiencing a dangerous situation that could result in an accident.

The danger/safety determination unit determines whether or not the vehicle is in a dilemma state based on the output of the collation unit, and at IIj1, if the determination result indicates that the vehicle is in a dilemma state, it determines the safety of the vehicle's passage at the difference point. A signal for control is output. The signals for this safety control are:
It is possible to output control signals for various safety purposes, such as adjusting the time of yellow lights and all red lights at intersection traffic lights, displaying street change information on roadside lanes, and road-to-vehicle communication.

The passing/stopping calculation unit uses a tracking method based on the output of the vehicle detection sensor, which also has a measurement castle within the intersection on the extension of the lane to be measured, and the position/speed calculation unit calculates the presence distribution. For each vehicle whose speed distribution was calculated,
Or did you pass through the intersection between yellow light time and full red time?
Alternatively, it is detected as a stop or pass/stop data.

This passing/stopping data and the position and velocity data obtained by the position/velocity calculation unit are sent to the learning unit, and this learning unit processes them within the update time interval of the dilemma state determination data file. In addition to collecting data,
Create a new dilemma state judgment data table using heuristic parametric learning or non-parametric learning, and send it to the matching unit.
The data table is updated.

To explain this more specifically, if the above passing/stopping +h data and position and speed data are collected within an appropriately set update time interval of the data table, m 2 t:
It is possible to obtain results as shown in J. In the speed-distance state diagram shown in the figure, the vertical axis shows the speed of the vehicle at the time the yellow light was generated, and the horizontal axis shows the position of the vehicle at the time the yellow light occurred, as a distance from the stop line. .

According to the figure, when a yellow light starts, vehicles that are far from the stop line and are slow will often stop, while vehicles that are close to the stop line and are fast will often pass, and those in between will stop and pass. However, this mixed part is a dilemma and represents a dangerous area.

7L The learning described above corresponds to determining this dangerous area, so the learning method is the simplest one, as long as it appropriately grasps the characteristics of this data and determines the dangerous area. The method is to locate the area surrounded by the four-tree straight line ~L4 in the diagram, that is, the vehicle that stopped at the position closest to the stop line when the yellow light started, and conversely, the vehicle that stopped at the position that was far away from the stop line. A straight line pointing upward to the right connects two pieces of data selected so as to envelop a mixed area from among the data for IItff4, which is located between the passing vehicle and has stopped at a high speed, and The area surrounded by a straight line pointing upward to the right connecting two pieces of data selected from among the data on slow or passing vehicles so as to enclose the mixed area can be defined as a dangerous area in a dilemma state.

Further, FIG. 3 shows a mixed area, that is, a dilemma area, derived from the overlap between the passing area and the stop F area obtained by the 6-fold calculation method.

In this way, from the collected data, through self-learning of dilemma situations that change with environmental and traffic conditions,
By creating a new cheater table for dilemma state determination and updating the data table in the collation section, it is possible to always apply optimal safety control to vehicles entering intersections.

Therefore, it is possible to detect in real time whether a vehicle entering a signalized intersection is dangerous or not, without sending the measurement results to a control center each time. In response to changes in vehicle traffic and traffic conditions, the vehicle will have intelligent functions that allow it to judge whether a vehicle is entering an intersection or is in a dangerous situation. In addition, the above-mentioned position/velocity calculation section, passing/stopping @n8
, the learning section, the collation section, and the danger/safety determination section can be easily realized by software in a computer.

[Effects of the Invention] According to the traffic flow sensor of the present invention detailed above, a dilemma situation that changes with environmental conditions and traffic conditions! A traffic flow sensor uses self-learning to detect ET, and thereby detects in real time whether a vehicle entering a signalized intersection is dangerous or not, without sending the measurement results to a control center etc. By applying online, real-time safety control, we can expect a significant reduction in the number of traffic stops.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a traffic flow sensor according to the present invention;
Figure 2 is a diagram of the dilemma region shown by four decision surfaces on the speed-distance state diagram of the vehicle at the inlet of the intersection.
FIG. 3 is a diagram of the dilemma area obtained by the 6-fold calculation method. Figure 1

Claims (1)

[Scope of Claims] 1. The road surface at the entrance of a signalized intersection is the object of measurement, and the position and speed of a vehicle passing there can be measured, and at the same time, it is possible to detect whether or not the vehicle has passed through the intersection. The vehicle detection sensor receives the signal at the start of the yellow light, and calculates and outputs the position and speed of the vehicle at the entrance of the intersection at the start of the yellow light from the output of the vehicle detection sensor.
Equipped with a speed calculation unit and a data table for determining dilemma status, it is equipped with
a verification unit that compares the output of the speed calculation unit with the data table; and a verification unit that determines whether the vehicle is in a dilemma state from the output of the verification unit and issues a signal for safe control of vehicle traffic at the intersection. Based on the output of the output danger/safety determination unit and the vehicle detection sensor, a tracking method is used to derive whether or not the detected vehicle has passed through the intersection.
A passing/stopping calculation unit calculates and outputs passing/stopping data, and collects the above passing/stopping data and the position and speed data obtained in the position/velocity calculation unit at appropriate update time intervals, a learning section that creates a new dilemma state determination data table through self-learning of dilemma states that change with traffic conditions and updates the data table of the collation section. sensor.