WO2021044707A1

WO2021044707A1 - Surroundings observation system, surroundings observation program, and surroundings observation method

Info

Publication number: WO2021044707A1
Application number: PCT/JP2020/024144
Authority: WO
Inventors: 幸彦小野
Original assignee: 株式会社日立製作所
Priority date: 2019-09-02
Filing date: 2020-06-19
Publication date: 2021-03-11
Also published as: JP2021037811A; JP7227879B2

Abstract

The purpose of the present invention is to provide a technique for increasing the precision of detecting a vehicle position. For this purpose, one typical surroundings observation system of the present invention comprises a surrounding environment observation unit and a position estimation unit. The surrounding environment observation unit observes a physical object group in the surroundings from a vehicle traveling on a railroad. The position estimation unit stores map data created by observing the physical object group in advance. Furthermore, the position estimation unit acquires, at a position limited to the railroad, a correlation of the physical object group between the result of the observation made by the surrounding environment observation unit and the map data, and estimates the vehicle position on the basis of the correlation.

Description

Peripheral observation system, peripheral observation program and peripheral observation method

The present invention relates to a peripheral observation system, a peripheral observation program, and a peripheral observation method.

In the technical field of railways, technology for finding obstacles in front of vehicles is known in order to improve safety. For example, in Patent Document 1, "an obstacle is detected by using a short-distance photographing unit for photographing a short distance provided in a vehicle, a long-distance photographing unit for photographing a long distance, and a LIDAR for irradiating a short distance. The technology to the effect of "doing" is disclosed.

With this type of obstacle detection technology, it was technically difficult to distinguish passengers who are safely waiting on the station platform and maintenance personnel who perform maintenance work safely near the track from obstacles.
Therefore, every time a non-obstacle is erroneously detected, the vehicle may be stopped in an emergency, which may hinder railway operation.

As a technique for preventing this kind of false detection, for example, Patent Document 2 states that "the vehicle position of a train is determined based on the correlation between a base video image of the front of a train taken in advance and a reference image of the front of a running train. Detect. A virtual building limit frame is preset for the base video image. The difference between the base video image and the reference video image taken at the same vehicle position at different dates and times is detected only within the virtual building limit frame. Detects only obstacles within the virtual building limit frame. "

Japanese Unexamined Patent Publication No. 2016-08183 JP-A-2017-001638

In a system that detects the vehicle position of a train as in Patent Document 2, it is desired to further improve the detection accuracy of the vehicle position.

Therefore, an object of the present invention is to provide a technique for peripheral observation that enhances the detection accuracy of the vehicle position.

In order to solve the above problems, one of the representative peripheral observation systems of the present invention includes a peripheral environment observation unit and a position estimation unit. The surrounding environment observation unit observes the position of a group of surrounding objects from the position of the vehicle traveling on the track and outputs the observation result. The position estimation unit stores map data created by observing the position of the object group in advance, and the observation result is in a state where the position of the vehicle in the observation result is limited to the track of the map data. The correlation between the vehicle and the map data is calculated, and the vehicle position is estimated based on the correlation.

The present invention provides a technique for peripheral observation that enhances the accuracy of detecting the vehicle position.

Issues, configurations and effects other than those described above will be clarified by the explanation of the following embodiments.

FIG. 1 is a diagram showing a configuration of an orbital transportation system (including a peripheral observation system). FIG. 2 is a diagram showing a configuration of a self-position estimation system. FIG. 3 is a flowchart showing the overall operation of the peripheral observation system. FIG. 4 is a diagram showing an example of a sensor for observing the surrounding environment. FIG. 5 is a diagram showing a lateral boundary of the monitoring area when traveling on a turning track. FIG. 6 is a flowchart showing the operation of the self-position estimation system. FIG. 7 is a diagram showing an example of the surrounding environment of the traveling vehicle. FIG. 8 is a diagram illustrating the detection of rail trajectories by the self-position estimation system. FIG. 9 is a diagram showing an example of surrounding environment observation data. FIG. 10 is a diagram for explaining the posture estimation of the vehicle by the self-position estimation system. FIG. 11 is a flowchart showing the operation of the vehicle driving control unit. FIG. 12 is a diagram showing an example of a monitoring area in a station section. FIG. 13 is a diagram showing an example of a monitoring area in the maintenance work section. FIG. 14 is a diagram showing scan matching (during scanning) in the case of an automobile. FIG. 15 is a diagram showing scan matching (scan completion) in the case of an automobile. FIG. 16 is a diagram showing scan matching (during scanning) of the embodiment. FIG. 17 is a diagram showing scan matching (scan completion) of the embodiment.

Hereinafter, embodiments will be described with reference to the drawings.

<< Configuration of Example 1 >>
FIG. 1 is a diagram showing a configuration of a railway transportation system 100 (including a peripheral observation system 100A) of the first embodiment.

The railway transportation system 100 includes a vehicle 102 and a peripheral observation system 100A.

Vehicle 102 is a vehicle that transports passengers and freight that travel along the track. The vehicle 102 is composed of a vehicle driving control unit 105 and a vehicle drive unit 106. Obstacle information 161 is given to the vehicle 102 from the peripheral observation system 100A.

The vehicle driving control unit 105 has an internal function of detecting the position and speed of the vehicle 102. The vehicle driving control unit 105 generates a drive command 142 so that the position and speed follow the target traveling pattern. The target traveling pattern is based on a pattern based on the acceleration / deceleration of the vehicle 102 and the speed limit of the traveling section, which are known in advance. Then, the vehicle driving control unit 105 calculates the allowable maximum speed of the vehicle 102 from the position of the vehicle 102 and the maximum deceleration of the vehicle 102, and reflects it in the basic target traveling pattern. As such a vehicle operation control unit 105, an ATO device (automatic train operation device) can be mentioned as an example.

The vehicle drive unit 106 drives the vehicle 102 based on the given drive command 142. Examples of the specific device of the vehicle drive unit 106 include an inverter, a motor, a friction brake, and the like.

On the other hand, the peripheral observation system 100A includes a peripheral environment observation unit 107, a self-position estimation system 101, and an obstacle detection system 103.

The surrounding environment observation unit 107 is a sensor installed in front of the vehicle 102 to acquire the position, shape, color, reflection intensity, etc. of an object around the vehicle 102. For example, the surrounding environment observation unit 107 is a sensor such as a camera, a LIDAR (Laser Imaging Detection and Ringing), a laser radar, or a millimeter wave radar.

The self-position estimation system 101 acquires the surrounding environment observation data 151 from the surrounding environment observation unit 107, estimates the vehicle position and attitude of the vehicle 102, and outputs the position / attitude information 153.

The obstacle detection system 103 includes a detection range setting database 108, a monitoring area setting processing unit 109, a detection target information database 110, a side boundary monitoring unit 111, a front boundary monitoring unit 112, and an obstacle detection unit 113.

The detection range setting database 108 stores the detection range 154 of the monitoring area in association with the position / posture information 153 of the vehicle 102. The detection range 154 includes information on undetected areas such as the vicinity of the station platform and the area where maintenance work is performed, in addition to the range data based on the building limit near the track.

The monitoring area setting processing unit 109 acquires the detection range 154 corresponding to the position / attitude information 153 by inquiring the position / attitude information 153 to the detection range setting database 108. The monitoring area setting processing unit 109 sequentially determines the side boundary 155 of the monitoring area based on the latest detection range 154, and sets the side boundary monitoring unit 111. Further, the monitoring area setting processing unit 109 sequentially determines the front boundary 156 of the monitoring area based on the latest detection range 154, and sets the front boundary monitoring unit 112.

The detection target information database 110 records information such as the position and reflectance of the detection point (detection target other than obstacles) as the background observed in the monitoring area in association with the vehicle position and the monitoring area. The side boundary monitoring unit 111 and the front boundary monitoring unit 112 refer to the detection target information database 110 for the vehicle position and the monitoring area, and acquire

information

157 and 158 regarding the background detection points, respectively.

The side boundary monitoring unit 111 and the front boundary monitoring unit 112 detect obstacles in the area set on the side boundary and the front boundary of the monitoring area by using sensors such as a camera, LIDAR, laser radar, or millimeter wave radar. Has a function to do. Here, the side boundary monitoring unit 111 and the front boundary monitoring unit 112 may share the sensor of the surrounding environment observation unit 107.

The monitoring result 159 of the side boundary monitoring unit 111 is transmitted to the obstacle detection unit 113. Further, the monitoring result 160 of the front boundary monitoring unit 112 is transmitted to the obstacle detection unit 113.

The obstacle information 161 by the obstacle detection unit 113 is transmitted to the vehicle 102 as the output of the peripheral observation system 100A.

FIG. 2 is a diagram showing the configuration of the self-position estimation system 101.

In the figure, the self-position estimation system 101 includes an observation data selection processing unit 114, a vehicle attitude estimation processing unit 115, a peripheral environment data coordinate conversion processing unit 116, a peripheral environment map database 117, a 3D rail track database 118, and a scan matching self. The position estimation unit 119 is provided.

Such a peripheral observation system 100A is composed of one or more computer systems equipped with, for example, a CPU (Central Processing Unit) and a memory as hardware. When this hardware executes the peripheral observation program, various functions of the peripheral observation system 100A are realized. For some or all of this hardware, dedicated equipment, general-purpose machine learning machines, DSP (Digital Signal Processor), FPGA (Field-Programmable Gate Array), GPU (Graphics Processing Unit), PLD (programmable logic device) You may substitute with. Further, by arranging a part or all of the hardware in a cloud in a centralized or distributed manner on a server on an external network, a plurality of peripheral observation systems 100A may be jointly used via the network.

<< Overall operation of peripheral observation system 100A >>
FIG. 3 is a flowchart illustrating the overall operation of the peripheral observation system 100A.
In steps S201 to 205 shown in the figure, whether or not to give a stop instruction to the vehicle 102 is determined for each measurement cycle of the obstacle detection system 103.

Hereinafter, description will be given according to the step numbers shown in FIG.

Step S201: The monitoring area setting processing unit 109 in the obstacle detection system 103 acquires the position / attitude information 153 calculated by the self-position estimation system 101.

Step S202: The monitoring area setting processing unit 109 queries the detection range setting database 108 for the position / attitude information 153 to obtain the side boundary 155 and the front boundary 156 as the boundaries of the monitoring area for monitoring obstacles. For example, the building limits on the left and right of the track are set as the side boundary 155 of the obstacle monitoring area, and the stoptable distance of the vehicle is set as the front boundary 156 of the obstacle monitoring area.

Here, in order to quickly detect an obstacle that invades beyond the lateral boundary 155 and the front boundary 156 on the boundary, the

side boundaries

155a and 155b and the front boundary 156 shown in FIG. 4 have boundary lines having a predetermined width. A widened boundary detection area is set. This predetermined width takes into consideration the size and maximum moving speed of the obstacle expected to exceed the boundary and the sensing cycle of the side boundary monitoring unit 111 and the front boundary monitoring unit 112 (obstacle detection sensor). It is set to ensure a width that can be detected at least once when an object enters the boundary.

For example, in the station section, the predetermined width is set narrowly to several cm to several tens of cm (more specifically, 10 cm) assuming the size of passengers waiting at the platform and the maximum moving speed.
Further, for example, in a traveling section between stations (particularly near a railroad crossing), a predetermined width is set wide (for example, 1 m) assuming the size and maximum moving speed when a car or the like crosses. In this way, the predetermined width of the boundary detection region is sequentially changed according to the vehicle position of the vehicle 102.

Step S203: The side boundary monitoring unit 111 and the front boundary monitoring unit 112 monitor whether or not an obstacle has entered the orbit beyond the side boundary 155 and the front boundary 156, and monitor the monitoring results 159 and 160 as obstacles. It is given to the detection unit 113.

As shown in FIG. 4, the boundary detection regions of the left and right

lateral boundaries

155a and 155b can be rectangular regions having a width of several tens of centimeters and a depth of more than 100 m. In order to keep the rectangular area within the observation range, as the

detectors

201 and 202 of the lateral boundary monitoring unit 111, two multilayer LIDARs or the like installed at high positions on the front left and right of the vehicle 102 so as to face downward are used. As the

detectors

201 and 201, one or more stereo cameras, millimeter-wave radars, and laser rangefinders may be used. Further, the boundary detection region of the lateral boundary 155 may be scanned by attaching the sensor to the automatic pan head.

Further, for example, as the detector 203 of the front boundary monitoring unit 112, a monocular camera having a narrow angle of view, an infrared camera for a dark field, or a stereo camera in consideration of the fact that the boundary detection area of the front boundary 156 is far away. , Millimeter wave radar, multilayer LIDAR, laser rangefinder, etc. are used.

The detection results of these detectors 201 to 203 are compared with the background detection points (detection targets other than obstacles) recorded in the detection target information database 110. Based on this comparison, when any of the following conditions 1 to 3 is satisfied, it is determined that an obstacle exists in the boundary detection region.

(Condition 1) The detection point in the boundary detection area is not detected.

(Condition 2) The position of the detection point in the boundary detection area is different.

(Condition 3) The reflection intensity of the detection point in the boundary detection area is different.

Here, as the speed of the vehicle increases, the distance required for the transport vehicle to stop increases, and the front and side boundary detection areas expand far away. At this time, if the reflectance of the laser at a distant detection point (road surface, installed object, etc.) becomes extremely small, it is erroneously determined that an obstacle has invaded under Condition 1. In order to avoid such a misjudgment, the traveling speed of the vehicle must be suppressed.

Therefore, in order to avoid misjudgment without reducing the running speed, the following measures 1 and 2 are added.

(Countermeasure 1) Only the position of an existing object (rail, sign, etc.) with a reflectance of a certain value or higher in the boundary detection area is set as the detection point.

(Countermeasure 2) A marker having a reflectance of a certain value or higher is set in advance as a detection point in the boundary detection area.

In either case, the position of the detection target and its reflectance are recorded in the detection target information database 110 in advance, and the detection point is regarded as an obstacle only when the position of the detection point is included in the boundary detection area at the current vehicle position. Used to determine intrusion.

FIG. 5 shows a case where the multilayer LIDAR is used as the surrounding environment observation unit 107 and the detectors 201 to 203 as shown by a plurality of straight lines. By using the detection points on the boundary detection area that spans multiple layers, it is possible to determine the intrusion of obstacles even when the boundary is a curve.

In FIG. 5, the road surface detection points by each detection layer irradiated from the multilayer LIDAR are shown by dotted lines. The detection layers that pass over the

lateral boundaries

155a and 155b depend on the distance from the vehicle, and the detection points of those multiple layers are monitored to determine the intrusion of obstacles.

At this time, even if the detection point of LIDAR is on the boundary detection area, if the straight line (laser optical path) connecting the detection point and LIDAR passes outside the boundary detection area, it is not used for determining the intrusion of obstacles. .. This is to prevent false detection by an object outside the boundary detection area.

By using a plurality of sensors of different types as the detectors 201 to 203, the probability of detecting an obstacle may be increased and the detection rate of the obstacle may be increased. Further, the false detection rate may be lowered by using the logical product (AND) of the detected detection results.

When an obstacle enters the orbit beyond the side boundary 155 and the front boundary 156, the obstacle detection unit 113 shifts the operation to step S204. In other cases, the obstacle detection unit 113 shifts the operation to step S205.

Step S204: When it is determined in step S203 that an obstacle exists, the obstacle detection unit 113 creates obstacle information 161 in order to stop the vehicle 102.

Step S205: The obstacle detection unit 113 transmits the obstacle information 161 to the vehicle 102.

The above is a description of the overall operation of the peripheral observation system 100A.
Next, the operation of the self-position estimation system 101 will be described in detail.

<< Operation of self-position estimation system 101 >>
FIG. 6 is a flowchart illustrating the operation of the self-position estimation system 101.

Steps S401 to 408 shown in the figure are operations for estimating the vehicle position of the vehicle 102, and are repeatedly executed every measurement cycle of the obstacle detection system 103.

Step S401: The observation data selection processing unit 114 acquires the peripheral environment observation data 151 observed by the peripheral environment observation unit 107. For example, when the surrounding environment of the vehicle 102 shown in FIG. 7 is observed by the multilayer LIDAR, rail observation data 162a to 162af and the like are acquired as three-dimensional point group data as shown in FIG.

Step S402: The observation data selection processing unit 114 sorts the acquired surrounding environment observation data into the rail observation data 162a to 162 on the orbit shown in FIG. 9 and the peripheral observation data 163 other than the orbit.

Such rail observation data 162a to 162 are selected by utilizing the shape and reflectance of the rail observation data and the fact that the rail detection data forms one raceway surface (plane or curved surface).

Step S403: The vehicle attitude estimation processing unit 115 determines the temporary position of the vehicle 102 on the rail track in the map data of the surrounding environment map database 117 and the 3D rail track database 118. This temporary position is determined based on the communication between the rail ground element and the vehicle 102, GPS information, the cumulative speed of the vehicle 102, and the like.

Step S404: The vehicle attitude estimation processing unit 115 acquires rail position information (three-dimensional point cloud data defining the rail surface) at the temporary position by inquiring the temporary position on the rail track to the 3D rail track database 118. ..

The vehicle attitude estimation processing unit 115 geometrically calculates the surface R formed by the rail surface obtained from the rail observation data 162 shown in FIG. 10 and the rail position information acquired from the 3D rail track database 118 to obtain the attitude of the vehicle 102. Estimate the information.

Step S405: Since the peripheral environment observation data of the peripheral environment observation unit 107 is observed by the sensor mounted on the vehicle 102, it becomes an observed value in the _{vehicle coordinate system Σ T fixed to the vehicle 102.}

On the other hand, the map data and the 3D rail track data in the _{external coordinate system Σ O} are recorded in the surrounding environment map database 117 and the 3D rail track database 118.

The surrounding environment data coordinate conversion processing unit 116 obtains the surrounding environment observation data in the vehicle coordinate system fixed to the vehicle 102 based on the temporary position of the vehicle 102 on the rail track and the attitude information of the vehicle 102 estimated in step S404. The _{coordinates are converted from Σ T} to the external coordinate system Σ _O, and output as the surrounding environment data after the coordinate conversion.

Step S406: The scan matching self-position estimation unit 119 scan-matches the surrounding environment data after coordinate conversion with the map data recorded in the surrounding environment map database 117. This scan matching scans with the sensor position (vehicle position) in the surrounding environment data constrained on the rail track of the map data. The scan range in this case is a width that can maintain the validity of the posture estimation in step S404 and the coordinate transformation in step S405.
In this scan range, the scan position where the residual between the surrounding environment data and the map data (such as the sum of the absolute values of the differences between the two data) is the minimum residual is searched.

Step S407: The scan matching self-position estimation unit 119 determines whether or not the minimum residual obtained in scan matching is smaller than the permissible value. This permissible value is appropriately set based on the accuracy required for the vehicle position.

Here, when the minimum residual exceeds the permissible value, the scan matching self-position estimation unit 119 shifts the operation to step S408.

If the minimum residual is within the permissible value, the scan matching self-position estimation unit 119 shifts the operation to step S409.

Step S408: The scan matching self-position estimation unit 119 displaces the temporary position along the rail track by the distance. This step distance is a distance that exceeds the range scanned in step S406. Further, the displacement direction of the temporary position (positive or negative of the step distance) is determined in the direction in which the matching residual becomes smaller based on the change tendency of the residual during the scan matching in step S406.

After the temporary position is displaced in this way, the scan matching self-position estimation unit 119 returns the operation to step S404.

Step S409: By repeating steps S404 to S408, the minimum residual of scan matching is reduced to be equal to or less than the allowable value. The scan matching self-position estimation unit 119 estimates the scan position on the rail track that has obtained the minimum residual of the permissible value or less as the vehicle position of the vehicle 102.

<< Operation of vehicle driving control unit 105 >>
Next, the operation of the vehicle driving control unit 105 will be described.

FIG. 11 is a flowchart illustrating the operation of the vehicle driving control unit 105.
The operation shown in the figure is repeatedly executed at regular intervals.
Step S500: The vehicle operation control unit 105 acquires information on the location of the vehicle 102 based on the operation schedule of the vehicle 102 and the like.

Step S5011: The vehicle driving control unit 105 determines whether or not the vehicle 102 is stopped at the station. The determination is made based on the location and speed of the vehicle 102. For example, if the vehicle 102 is near the station platform and the speed is zero, it is determined that the vehicle is stopped at the station. When it is determined that the vehicle is stopped at the station, the vehicle operation control unit 105 shifts the operation to step S502. In other cases, the vehicle driving control unit 105 shifts the operation to step S511.

Step S502: The vehicle operation control unit 105 acquires information on the scheduled departure time of the vehicle 102 based on the operation schedule of the vehicle 102.

Step S503: The vehicle operation control unit 105 determines whether or not the current time has reached the scheduled departure time. If the current time has not reached the scheduled departure time, the vehicle operation control unit 105 exits this processing flow. When the current time reaches the scheduled departure time, the vehicle operation control unit 105 proceeds to step S504.

Step S504: The vehicle driving control unit 105 determines whether or not the vehicle 102 has completed the departure preparation. An example of preparation for departure is confirmation of the closed state of the vehicle door. If it is not completed, exit this processing flow. When the departure preparation is completed, the vehicle operation control unit 105 proceeds to step S505.

Step S505: The vehicle driving control unit 105 acquires obstacle information 161 from the obstacle detection system 103. In this case, as shown in FIG. 12, the obstacle monitoring area 500 is set to exclude the range of the station platform based on the highly accurate vehicle position by the self-position estimation system 101. By using a highly accurate vehicle position, a highly accurate setting that includes the inside of the white line of the station platform and the vicinity of the platform door (the area where the vehicle 102 is not safe because it passes nearby) in the obstacle monitoring area 500. Is fully possible. Further, based on the highly accurate vehicle position in the station section, it is possible to narrow the monitoring area 500 so that the vehicle on the adjacent rail (vehicle on another platform of the station) is not detected as an obstacle.

Step S506: The vehicle driving control unit 105 determines whether or not an obstacle exists from the obstacle information 161. If there are no obstacles, the vehicle driving control unit 105 proceeds to step S507. When an obstacle is present, the vehicle driving control unit 105 exits the main processing flow while generating or maintaining a stop command.

Step S507: The vehicle driving control unit 105 generates a drive command 142 and transmits it to the vehicle drive unit 106. Specifically, here, a powering command is transmitted to depart the station.

Step S508: The vehicle operation control unit 105 calculates the estimated time of arrival of the next station based on the timing when the vehicle 102 departs and the estimated time of travel between the stations to be traveled, and the operation management center of the vehicle 102 ( Send to the operation management system).

Next, processing when it is determined in step S501 that the vehicle 102 is not stopped at the station (steps S511 to S515) will be described.

Step S311: The vehicle driving control unit 105 acquires the obstacle information 161 in the monitoring area from the obstacle detection system 103. As shown in FIG. 13, the obstacle monitoring area 600 is widely set because it is traveling between stations. However, if there is a section under maintenance work between stations, the range where maintenance personnel can work safely should be excluded from the obstacle monitoring area 600 based on the highly accurate vehicle position by the self-position estimation system 101. Is set to.

Step S512: The vehicle driving control unit 105 determines whether or not braking of the vehicle 102 is necessary based on the obstacle information 161 in the monitoring area. If there are no obstacles and braking is not required, the vehicle driving control unit 105 proceeds to step S513. If there is an obstacle and braking is required, the vehicle driving control unit 105 proceeds to step S514.

Step S513: The vehicle driving control unit 105 generates a drive command 142 and transmits it to the vehicle drive unit 106. Specifically, here, a drive command such as proportional control is transmitted so that the speed of the vehicle 102 becomes a predetermined target speed. After transmission, the vehicle driving control unit 105 advances the operation to step S515.

Step S514: The vehicle driving control unit 105 transmits a braking command of the vehicle 102 to the vehicle driving unit 106 in response to the detection of an obstacle. Specifically, a braking command is generated to decelerate and stop the vehicle 102 at the maximum deceleration. After transmitting the braking command, the vehicle driving control unit 105 advances the operation to step S515.

Step S515: The vehicle operation control unit 105 estimates the time when the vehicle 102 arrives at the next station from the position, speed, and operation status at that time, and transmits the time to the operation management center (operation management system) of the vehicle 102.

The above is the operation explanation of the vehicle driving control unit 105.

<< Effects of Example 1 etc. >>
In the first embodiment, the matching (correlation) between the observation result of the surrounding environment observation unit 107 and the map data in the surrounding environment map database is scanned in a range along the orbit.

Normally, in the case of a vehicle that does not travel along a specific track such as an automobile, as shown in FIGS. 14 and 15, the surrounding environment data 168 and the map data 166 are correlated while repeating displacements in a plurality of directions. The position with the highest correlation value (FIG. 15) is obtained as the vehicle position. In this case, since the degree of freedom of scanning is high, it is easy to erroneously determine the position where the matching residual is the minimum as the vehicle position, and there is a problem that it is difficult to improve the detection accuracy of the vehicle position.

Furthermore, since scanning in multiple directions is required, there is also a problem that the processing load required for scan matching is heavy and the processing takes time.

However, in the vehicle position estimation process in the first embodiment, the matching (correlation) between the surrounding environment data 168 and the map data 166 is the highest while scanning one-dimensionally while being constrained to the track L as shown in FIG. The scan position (see FIG. 17) is estimated as the vehicle position. Therefore, since the scanning range is limited to the track L, it is possible to improve the detection accuracy of the vehicle position without erroneously determining the position deviating from the track L as the vehicle position.

Further, in the first embodiment, since the scan range is limited by the trajectory L, the processing load of scan matching is lightened, and the time required for processing can be shortened.

Further, in the first embodiment, the posture of the vehicle 102 with respect to the track is obtained based on the observation result of the track by the surrounding environment observation unit 107. Even for the vehicle 102 traveling on the track, the posture of the vehicle 102 with respect to the track fluctuates from moment to moment due to the influence of acceleration / deceleration, curve traveling, and the like. In the first embodiment, it is possible to acquire information on the posture of the vehicle 102 that fluctuates in this way.

Further, in the first embodiment, the observation result of the surrounding environment observation unit 107 is coordinate-transformed based on the posture of the vehicle 102. By this coordinate transformation, the tilt fluctuation of the observation result of the surrounding environment observation unit 107 due to the posture of the vehicle 102 is correctly calibrated. Therefore, the matching accuracy between the observation result after the coordinate conversion and the map data becomes high, and it becomes possible to further improve the detection accuracy of the vehicle position.

Further, in the first embodiment, since the detection accuracy of the vehicle position is high, the monitoring area of the obstacle detection system 103 can be changed at an appropriate timing according to the vehicle position. Therefore, it is possible to reliably detect only obstacles in the monitoring area to be monitored. In addition, an object located outside the monitoring area that should not be monitored will not be erroneously detected as an obstacle.

For example, in the station section, by narrowing the monitoring area, passengers waiting at a safe position on the station platform (or outside the white line of the station platform, etc.) will not be mistakenly detected as obstacles.

Also, for example, in the maintenance work section, by narrowing the monitoring area, maintenance personnel who work safely will not be mistakenly detected as obstacles.

Further, in the first embodiment, a boundary detection area having a predetermined width is provided at the boundary of the monitoring area (left and right lateral boundary, front boundary, etc.) for the purpose of quickly and surely detecting an obstacle. In the first embodiment, since the detection accuracy of the vehicle position is high, it is possible to change the predetermined width of the boundary detection area at an appropriate timing according to the vehicle position.

For example, in the station section, by narrowing the predetermined width of the boundary detection area, it becomes possible to intensively detect passengers invading on the railroad track or in the white line of the station platform.

Further, for example, in the traveling section between stations, by expanding the predetermined width of the boundary detection area, it is possible to reliably detect the fast movement of a relatively large moving object (automobile, etc.) invading the railroad track from a relatively wide field of view. Becomes possible.

The present invention is not limited to the above-described embodiment, and includes various modifications. For example, the above-described embodiment has been described in detail in order to explain the present invention in an easy-to-understand manner, and is not necessarily limited to the one including all the described configurations.

It is also possible to add / delete / replace a part of the configuration of the embodiment with another configuration.

For example, in the embodiment, the operation of intensively detecting an obstacle in the boundary detection area of the monitoring area is described, but the present invention is not limited to this. For example, obstacles may be detected in the entire area or subarea within the monitoring area.

100 ... Rail transport system, 100A ... Peripheral observation system, 101 ... Self-position estimation system, 102 ... Vehicle, 103 ... Obstacle detection system, 105 ... Vehicle operation control unit, 106 ... Vehicle drive unit, 107 ... Surrounding environment observation unit, 108 ... Detection range setting database, 109 ... Monitoring area setting processing unit, 110 ... Detection target information database, 111 ... Side boundary monitoring unit, 112 ... Front boundary monitoring unit, 113 ... Obstacle detection unit, 114 ... Observation data selection processing Unit, 115 ... Vehicle attitude estimation processing unit, 116 ... Surrounding environment data coordinate conversion processing unit, 117 ... Surrounding environment map database, 118 ... 3D rail track database, 119 ... Scan matching self-position estimation unit, 142 ... Drive command, 151 ... Surrounding environment observation data, 153 ... position / attitude information, 154 ... detection range, 155 ... side boundary, 156 ... front boundary, 159 ... monitoring result, 160 ... monitoring result, 161 ... obstacle information

Claims

The surrounding environment observation unit that observes the position of surrounding objects from the position of the vehicle traveling on the track and outputs the observation result,
The map data created by observing the position of the object group in advance is stored, and the observation result and the map data are combined with the position of the vehicle in the observation result limited to the track of the map data. A peripheral observation system including a position estimation unit that calculates the correlation between the above and estimates the vehicle position based on the correlation.
The peripheral observation system according to claim 1.
The surrounding environment observation unit observes the orbit and
The position estimation unit is a peripheral observation system characterized in that the posture of the vehicle with respect to the track is obtained based on the observation result of the track.
The peripheral observation system according to claim 2.
The position estimation unit performs coordinate conversion of the observation result of the surrounding environment observation unit based on the posture of the vehicle, and the correlation between the observation result after the coordinate conversion and the map data of the object group. A peripheral observation system characterized in that the vehicle position is estimated based on the correlation.
The peripheral observation system according to any one of claims 1 to 3.
An obstacle detection unit that detects obstacles existing in the monitoring area set for the orbit, and
A peripheral observation system including a boundary setting unit that changes the monitoring area based on the vehicle position estimated by the position estimation unit.
The peripheral observation system according to claim 4.
The monitoring area includes a boundary detection area having a predetermined width for detecting an obstacle at the boundary of the monitoring area.
The boundary setting unit is a peripheral observation system characterized in that the predetermined width of the boundary detection region is changed based on the vehicle position estimated by the position estimation unit.
The peripheral observation system according to any one of claims 4 to 5.
The peripheral observation system is characterized in that the monitoring area is defined by a lateral boundary set to the left and right of the orbit and a front boundary set to the front in the traveling direction of the orbit.
The peripheral observation system according to any one of claims 4 to 6.
The boundary setting unit is a peripheral observation system characterized in that the monitoring area in the station section is changed based on the positional relationship between the vehicle position estimated by the position estimation unit and the station section.
The peripheral observation system according to any one of claims 4 to 7.
The boundary setting unit changes the monitoring area in the maintenance work section based on the positional relationship between the vehicle position estimated by the position estimation unit and the maintenance work section of the track. system.
A peripheral observation program characterized in that a computer system functions as the peripheral observation system according to any one of claims 1 to 8.
Surrounding environment observation steps for observing surrounding objects from vehicles traveling in orbit,
The map data created by observing the object group in advance is stored, and the correlation between the observation result by the surrounding environment observation step and the map data is taken at a position limited to the orbit. , A peripheral observation method including a position estimation step for estimating a vehicle position based on the correlation.