JP7650653B2 - Road Traffic Control System - Google Patents

Description

The present invention relates to a road traffic control system that provides autonomous vehicles passing through intersections with a wide range of future prediction information that cannot be fully understood by the autonomous vehicles alone, so that the autonomous vehicles can travel safely.

Patent Document 1 describes a vehicle collision prevention device for preventing head-on collisions at intersections. This vehicle collision prevention device calculates the probability distribution of the intersection entry time of multiple vehicles entering the intersection based on the driving conditions of each vehicle, determines the probability of collision by comparing these probability distributions, and issues an alarm to avoid collisions.

Japanese Patent Application Publication No. 6-231396

At intersections, vehicles entering the intersection may stop, start, accelerate or decelerate due to changes in the color of traffic lights or the presence of pedestrians, and road conditions may change due to changes in weather such as rain or snow.

However, the technology in Patent Document 1 mentioned above does not take into account the effects of changes in the color of traffic lights, the presence or absence of pedestrians, or changes in road conditions, and calculates the collision probability only from the driving conditions based on the acceleration/deceleration and position of the vehicle, so it does not reflect changes in actual traffic conditions.

However, autonomous vehicles make various decisions based on the surrounding environment and driving conditions that can be recognized by the vehicle's sensors, so there is an issue that it is difficult for the systems installed in autonomous vehicles alone to detect changes in traffic conditions over a wide area and predict future trends.

The present invention was made in consideration of the above circumstances, and its purpose is to provide a road traffic control system that can make future predictions in line with changes in traffic conditions over a wide range that cannot be fully grasped by autonomous vehicles alone.

A road traffic control system according to one embodiment of the present invention is a system that predicts the future position of a moving body detected by a moving body sensor, determines the possibility of a collision with the moving body based on position information and route information received from an autonomous vehicle, and notifies the autonomous vehicle. The system is configured to predict the color of the signal lights when the autonomous vehicle approaches an intersection based on signal information from a traffic signal controller, correct the determination result of the possibility of collision based on the predicted signal information and environmental information including road surface conditions from an environmental sensor, and notify the collision risk by dividing it into levels. The determination of the possibility of collision is characterized in that a polygon with the maximum area is formed with vertices representing the future position of a specified object after a predetermined time and each point of the future position immediately before the future position, and a polygon with the maximum area is formed for all other objects with which the driving route may intersect, and if the polygons overlap, a collision is deemed to have occurred.

The road traffic control system of the present invention corrects the judgment of the possibility of a collision or contact accident between a moving object and an autonomous vehicle based on signal information from a traffic signal controller and environmental information including the road surface conditions at the intersection, making it possible to make future predictions based on changes in actual traffic conditions over a wide range of areas that cannot be grasped by autonomous vehicles alone.

1 is a block diagram showing a schematic configuration of a road traffic control system according to an embodiment of the present invention. 2 is a perspective view for explaining collision determination when a right-turning vehicle and an oncoming straight-moving vehicle approach each other at an intersection to which the road traffic control system shown in FIG. 1 is applied. FIG. FIG. 3 is a schematic plan view of the intersection shown in FIG. 2 . 3 is a schematic plan view for explaining a collision determination in a case where a right-turning vehicle stops temporarily within the intersection shown in FIG. 2. FIG. 2 is a perspective view for explaining collision determination between a left-turning or right-turning vehicle and a pedestrian or a bicycle crossing a crosswalk at an intersection to which the road traffic control system shown in FIG. 1 is applied. FIG. FIG. 13 is a diagram for explaining area ranges of straight line portions and curved line portions. 13 is a diagram for explaining a method for setting an area range of a curved line portion. FIG. FIG. 11 is a diagram for explaining determination of an area position. FIG. 13 is a diagram for explaining a unit position. FIG. 11 is a diagram for explaining the size of an object used in collision determination. FIG. 11 is a schematic diagram for explaining collision determination in detail. 2 is a block diagram showing an example of a configuration in which direct communication with vehicles is performed using an IP network in the road traffic control system shown in FIG. 1. FIG. 2 is a block diagram showing an example of a configuration in which communication with vehicles is performed via a server using an IP network in the road traffic control system shown in FIG. 1. FIG. FIG. 2 is a block diagram showing a configuration example in which a first (last) mile management system is applied to the road traffic control system shown in FIG. 1 .

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
Fig. 1 shows a schematic configuration of a road traffic control system according to an embodiment of the present invention. This system is composed of a roadside unit 1 installed near an intersection and an on-board device 2 installed in an autonomous vehicle traveling near the intersection.
The roadside unit 1 includes a control module 11, a traffic signal controller 12, a multi-sensor module (mobile sensor) 13, an environmental sensor 14, and a wireless device 15.

The in-vehicle equipment 2 includes various sensors, a vehicle control device, an automatic driving (or driving assistance) device, and a wireless device 17 mounted on an automatic driving vehicle 16. The vehicle control device is composed of an engine control device that controls the engine, a steering control device that controls the steering, and a brake control device that controls the brakes. The automatic driving device calculates the current position and driving route from the position information and route information obtained from the various sensors, and controls the vehicle control device based on the calculation results to drive the vehicle.

The control module 11 receives signal information SA1 from the traffic signal controller 12, moving object information SA2 from the multi-sensor module 13, and environmental information SA3 from the environmental sensor 14. In addition, the control module 11 receives driving information SA4 transmitted from the wireless device 17 of the autonomous vehicle 16 via the wireless device 15.

The control module 11 is configured to predict the future positions of moving objects such as non-self-driving vehicles, bicycles, and pedestrians using the moving object information SA2 from the multi-sensor module 13, and to determine the possibility of a collision with the moving object based on the driving information SA4 (identification information SA4a, position information SA4b, route information SA4c, etc.) received from the self-driving vehicle 16. When making this determination, the determination result is corrected based on the signal information SA1 from the traffic signal controller 12 and the environmental information SA3 including the road surface conditions from the environmental sensor 14. Then, the wireless device 15 transmits roadside information SA5 (identification information SA4a, signal information SA1, danger information SA5a, etc.) to the self-driving vehicle 16.

The traffic signal controller 12 controls the light color and signal switching of the traffic lights, and outputs signal information SA1, such as the light color and the time (in seconds) until the signal switches, to the control module 11. The control module 11 transmits the signal information SA1 from the wireless device 15 before the autonomous vehicle 16 enters the intersection. By receiving the light color and the timing of the signal switch in advance, the autonomous vehicle 16 predicts the light color of the traffic lights when it approaches the intersection.

The multi-sensor module 13 detects vehicles (non-self-driving vehicles) within the intersection, for example, up to a distance of about 150 m around the intersection, as well as bicycles and pedestrians around the crosswalk and intersection (as far away as possible). In this example, the multi-sensor module 13 is an imaging sensor, and includes a vehicle light camera 21 and a pedestrian light camera 22 that function as moving object sensors. The vehicle light camera 21 is attached, for example, to a high position on a traffic light pole, and captures an overhead view of vehicles passing through the intersection. The pedestrian light camera 22 is attached, for example, near the pedestrian signal on a traffic light pole, and captures pedestrians (including bicycles) crossing the crosswalk.

The images captured by these cameras 21, 22 are processed by an image processor 23 to detect vehicles and pedestrians around the intersection. Then, vehicle information SA2a such as the position (longitude and latitude), speed, and moving direction of the detected vehicles, and pedestrian information SA2b such as the position (longitude and latitude), speed, and moving direction of the detected pedestrians are input to the control module 11 as moving object information SA2.
Although an example has been described in which an imaging sensor is used as the multi-sensor module 13, LiDAR (light detection and ranging) may be used, which measures the distance and direction to an object by irradiating the object with laser light and measuring the time it takes for the light to bounce back after hitting the object. By using LiDAR, the distance, shape, and positional relationship of vehicles, bicycles, pedestrians, etc. can be grasped in three dimensions.

The environmental sensor 14 is composed of a weather sensor 14a that detects the weather, and a road surface monitoring sensor 14b that monitors road surface conditions. The weather sensor 14a measures weather conditions such as wind direction, wind speed, temperature, humidity, air pressure, rainfall, and snowfall, and outputs meteorological information SA3a. The road surface monitoring sensor 14b also monitors road surface conditions such as wetness, snow accumulation, and freezing, and outputs road surface information SA3b. Based on this meteorological information SA3a and road surface information SA3b, the control module 11 judges the surrounding environment of the intersection.

The control module 11 not only determines whether the position and driving route of the autonomous vehicle 16 match the positions, speeds, and moving directions of detected vehicles and pedestrians, indicating the possibility of a collision, but also predicts the color of the traffic lights when the autonomous vehicle 16 approaches an intersection, and determines the risk of a collision based on the road surface conditions (slip-friction coefficient of the road surface) at the intersection, and sets the danger information SA5a to be sent to the autonomous vehicle 16. If the danger information SA5a has already been set and the surrounding environment has changed, the danger information SA5a is reset, in other words, the level ranking is changed.

Specifically, when the control module 11 receives identification information such as a vehicle ID, location information such as longitude, latitude, and speed, and route information such as the lane to be traveled, movement direction, and speed, transmitted from the autonomous vehicle 16 via the wireless device 17, using the wireless device 15, it creates a surrounding area map from the vehicle information SA2a and pedestrian information SA2b, and identifies the autonomous vehicle 16 from this surrounding area map. It also predicts the future positions of vehicles and pedestrians, and determines whether there is a danger if the autonomous vehicle 16 travels (whether there is a possibility of collision or contact with other vehicles or pedestrians).

In this determination, the road surface slip friction coefficient is calculated based on the environmental information obtained from the environmental sensor 14, and the braking distance taking into account the friction coefficient is calculated, for example, according to the following formula, and used to predict the autonomous vehicle 16 and the vehicle's future position.
Braking distance = speed before braking (km/h) squared ÷ (254 x coefficient of friction)
If it is determined that there is a danger, the degree of collision risk is classified into levels to determine the control required for avoidance, and the result of the determination, danger information SA5a, is transmitted from the wireless device 15 to the automatically driven vehicle 16.

On the side of the autonomous vehicle 16, for example, the traffic light information SA1 is used to predict the light color and the timing (time) of the next traffic light change, and control is performed such as maintaining speed, decelerating, stopping, starting, etc. That is, when the light color is green at the time when it is predicted to approach the intersection, the future position is predicted to pass through the intersection, and when it turns yellow or red, the future position is predicted to stop. When the light color is red at the time when it is predicted to approach the intersection, the future position is predicted to stop.
In addition, depending on the level of the danger information SA5a, braking operations and engine output are suppressed, and control such as deceleration and stopping is performed to avoid collisions or contact.

The collision risk level is classified into a plurality of levels according to the deceleration required for the autonomous vehicle to avoid the risk, for example, a first level (normal deceleration), a second level (maximum deceleration), a third level (emergency braking), and a fourth level (damage mitigation braking).
The first stage is a braking operation that will cause a collision if moving at the current speed, but will allow the vehicle to stop just before the predicted collision area if it decelerates at 0.2 G. It is assumed that the vehicle will decelerate (0.2 G) to avoid danger.
The second stage is a braking operation that will cause a collision if moving at the current speed, but will allow the vehicle to stop just before the predicted collision area if it decelerates at 0.3 G. It is assumed that the vehicle will decelerate (0.3 G) to avoid danger.

The third stage is a braking operation that will cause a collision if moving at the current speed, but will allow the vehicle to stop just before the predicted collision area if it slows down by 0.5 G. It is assumed that the vehicle will suddenly decelerate (0.5 G or more) to avoid danger.
The fourth stage is when the vehicle cannot stop before the predicted collision area even if it decelerates at 0.5G. Although the danger cannot be avoided, it is assumed that the vehicle will decelerate as much as possible to mitigate damage.

In this way, according to the road traffic control system of the present invention, the determination of the possibility of a collision or contact accident between an autonomous vehicle and a non-autonomous vehicle or pedestrian is corrected based on signal information from the traffic signal controller and environmental information including the road surface conditions at the intersection, making it possible to make future predictions that are in line with changes in actual traffic conditions over a wide range of areas that cannot be grasped by autonomous vehicles alone.
Furthermore, by dividing the degree of danger into multiple levels and notifying the autonomous vehicle of each level, it becomes possible for the vehicle to decelerate according to the degree of danger, rather than simply applying the brakes to bring the vehicle to a stop.

In FIG. 1, the control module 11 and the wireless device 15 may be built into a housing 18 of the traffic signal controller 12 as shown by a dashed line, or may be mounted externally on a signal pole.
For the sake of simplicity, the roadside unit 1 is described as being provided with one each of the control module 11, the traffic signal controller 12, the multi-sensor module 13, the environmental sensor 14, and the wireless device 15, but it is preferable to provide a plurality of multi-sensor modules 13 so as to be able to capture a wide range of images around the intersection. A plurality of environmental sensors 14 may also be provided as necessary, or it may be just the weather sensor 14a or the road surface monitoring sensor 14b.

Next, a case where danger information (here, oncoming vehicle approach information) is generated for each autonomous vehicle 16 in the above-mentioned road traffic control system will be described with reference to Figs. 2 and 3. Fig. 2 is a perspective view for explaining collision determination when a right-turning vehicle approaches an oncoming straight-moving vehicle at an intersection to which the road traffic control system shown in Fig. 1 is applied, and Fig. 3 is a schematic plan view of the intersection shown in Fig. 2.

Figure 2 shows a situation in which a right-turning vehicle (autonomous vehicle 16) and an oncoming straight-moving vehicle (non-autonomous vehicle 33) enter an intersection after a predetermined time, and the estimated positions of vehicles 16 and 33 after the predetermined time are shown by dashed lines (vehicles 16A and 33A).
The roadside unit 1 is attached to a signal pole 31, and receives driving information SA4, such as identification information SA4a, position information SA4b, and route information SA4c, transmitted from a wireless device 17 of an autonomous vehicle 16, with a wireless device 15 built into the roadside unit 1. In addition, a vehicle light camera 21 attached to a signal pole 32 captures an image of a vehicle (non-autonomous vehicle) 33 traveling in a hatched area AA, and an image processing unit 23 processes the image to detect the vehicle 33.

As in FIG. 2, vehicles 16A and 16B shown by dashed lines in FIG. 3 are the future positions of autonomous vehicle 16 a predetermined time later, and vehicles 33A and 33B shown by dashed lines are the future positions of vehicle 33 a predetermined time later. When autonomous vehicle 16 and vehicle 33 proceed along the dashed arrows, the future positions of vehicles 16B and 33B overlap within the intersection, and control module 11 determines that autonomous vehicle 16 and vehicle 33 will collide. Therefore, control module 11 determines the control required to avoid a collision, and wirelessly transmits risk information SA5a, which is divided into multiple risk levels, to autonomous vehicle 16.

Figure 4 is intended to explain collision judgment when a right-turning vehicle 16 (indicated by vehicle 16C in dashed lines) stops at the intersection shown in Figure 2, and judges whether or not an autonomous vehicle 16 traveling on a right-turn path 34 will collide (or come into contact) with a vehicle 33C traveling on a straight path 36 if it starts moving forward when stopped at a right-turn waiting point 35. That is, when the vehicle 16 starts moving, if an oncoming straight-moving vehicle 33C is present in the area AB where the right-turn path 34 and the straight path 36 intersect, as indicated by the arrow, between the right-turn waiting point 35 and the intersection passing point 37 (for example, 6 seconds from the present time), it is judged that a right turn is dangerous, and if not, that a right turn is permitted.

Even when making this risk assessment, by taking into consideration signal information SA1 from the traffic signal controller 12 and environmental information SA3 including the road surface conditions at the intersection, future predictions can be made in line with changes in actual traffic conditions over a wide range that cannot be grasped by the autonomous vehicle 16 alone.
Note that the time from the right turn waiting point 35 to passing the intersection passing point 37 is set to 6 seconds, however, this will of course vary depending on the width of the intersection, the acceleration performance of the autonomous vehicle 16, the speed of the oncoming straight-moving vehicle 33, etc., and it is advisable to set this time taking these factors into consideration.

In this way, the roadside unit 1 monitors the possibility of collision or contact between the autonomous vehicle 16 and the non-autonomous vehicle 33, and if it determines that there is a danger, it wirelessly transmits danger information SA5a that classifies the level of danger to the autonomous vehicle 16, making it possible to make future predictions in line with changes in traffic conditions over a wide range of areas that cannot be grasped by the autonomous vehicle 16 alone.

Next, the function of generating risk information for each intersection will be described with reference to Fig. 5. Fig. 5 is intended to explain collision judgment between a left-turning or right-turning vehicle and a pedestrian or bicycle crossing a pedestrian crossing at an intersection to which the road traffic control system shown in Fig. 1 is applied.
The roadside unit 1 is attached to a signal pole 31, and receives driving information SA4, such as identification information SA4a, position information SA4b, and route information SA4c, transmitted from the wireless devices 17 of the autonomous vehicles 40, 41, with the wireless devices 15. In addition, the pedestrian light camera 22 photographs pedestrians and bicycles on the crosswalk (hatched area AC). Although not shown, a pedestrian light camera is also provided on the opposite crosswalk across the intersection to photograph pedestrians and bicycles.

Here, a state in which a left-turning vehicle (automated vehicle) 40 and a right-turning vehicle (automated vehicle) 41 are about to enter an intersection at almost the same time is shown as an example. When the left-turning vehicle 40 and the right-turning vehicle 41 approach the intersection, the control module 11 receives identification information SA4a such as a vehicle ID, location information SA4b such as longitude, latitude, and speed, and route information SA4c such as the lane to travel, the direction of movement, and the speed, transmitted from the automatically-driven vehicles 40 and 41. In addition, a surrounding map is created from the vehicle information SA2a and the pedestrian information SA2b, and the automatically-driven vehicles 40 and 41 are identified from this surrounding map. Furthermore, the future positions of non-automated vehicles, pedestrians, and bicycles are predicted, and it is determined whether there is a danger when the automatically-driven vehicles 40 and 41 travel. At this time, it also detects pedestrians and bicycles ignoring traffic signals, and if it is determined that there is a danger, it determines the control required for avoidance, and transmits danger information SA5a (pedestrian, etc. warning information) in which the collision risk is classified into levels to the automatically-driven vehicles 40 and 41.

On the side of the autonomous vehicles 40, 41, for example, the light color and the timing of the next signal change are predicted using the traffic light information SA1, and control is performed such as starting, decelerating, and stopping. This improves the accuracy of future position predictions when approaching an intersection. In addition, braking operations and engine output are suppressed according to the level of the danger information SA5a, and control such as deceleration and stopping is performed. After the autonomous vehicle 40 passes through the intersection, if it is determined that there are no pedestrians or bicycles on the crosswalk (area AC), the autonomous vehicle 41 passes through the intersection.

In FIG. 5, for the sake of simplicity, the case of detecting pedestrians and bicycles on a crosswalk has been described. However, for example, multiple pedestrian light cameras 22 may be provided so as to detect pedestrians and bicycles around the intersection (as far away as possible).
Furthermore, by combining it with the vehicle lighting camera 21 shown in FIG. 2, moving objects can be detected over a wide area around the intersection, and the moving speed and direction can be accurately determined from multiple images.

Next, we will explain in detail how to estimate the future position of a moving body (object). This system predicts the future position of an object, such as a vehicle or pedestrian, from vector information input from the image processing unit 23, for example an AI image processing unit, and determines whether or not the object is a danger to an autonomous vehicle. When predicting the future position, the object is basically treated as performing actions that are considered appropriate given the surrounding circumstances.

However, if the object's current vector information deviates from what is considered to be a valid action, the object will take an action that is inferred from the current vector information. For example, when turning left and a pedestrian is moving on a crosswalk, the vehicle object will stop before the crosswalk, but if the vehicle object is moving on a vector that does not allow it to stop before the crosswalk, it will not stop before the crosswalk and will continue moving on the current vector.

The future position of each object is estimated based on the shape of the surrounding roads, the current position and speed of each object, etc.
Next, a process flow for estimating a future position of an object, area/route setting, area determination, route determination, area position determination, light color determination, behavior determination, object collision determination, and object avoidance behavior will be described.

<Processing flow>
As an example of the process flow for estimating future positions, the future position of each object is calculated at 0.2 second intervals for up to 15 seconds (specified by a program constant) into the future, starting from the current position. Each object basically moves along a route corresponding to the area in which it exists, accelerating and decelerating depending on the surrounding conditions (other objects, traffic lights, areas) and determining its final destination. The process is then repeated until it is time to calculate the future position, thereby calculating the future position.

<Area/Route Settings>
An area is defined as a division of a lane or the like into multiple sections. For straight lines, an area is defined by two points (the start and end points), and for curved lines, by four points (the start and end points and the control point of a cubic Bezier curve), and the area width.
Furthermore, a route is usually defined as the path an object travels. A route is defined as a collection of multiple contiguous areas. A single block is defined as one area, not multiple areas. However, the starting point of a branch may exist across multiple areas. Areas are basically represented by lines and points. This method has the advantage that paths such as right and left turns can be represented more naturally than the area method, and the length of the path itself can be obtained in the form of an approximate value.

As shown by the dashed lines in Figures 6(a) and (b), the range of the area is calculated by calculating lines 101, 102 or 111, 112 parallel to the input route information value (represented by a straight line 100 or a curve 110), and the width of the calculated lines is set as the lane width 103, 104 or 113, 114. In other words, areas 105 and 115 are defined as the area surrounded by two parallel lines and a straight line of the length of the area width drawn in the normal direction of the start and end points.

As shown in Figure 7, in a Bezier curve defined by a curved portion, normal lines are drawn perpendicular to the straight lines connecting the start point and the start control point, and the end point and the end control point. When calculating two lines parallel to an area, the start and end points are determined, and then the control points are determined according to the ratio to the original start and end points. After that, the points through which the parallel lines pass are calculated according to the Bezier curve formula.

In the formula for a Bezier curve, when the start point is ( _x1 , _y1 ), the start control point is ( _x2 , _y2 ), the end control point is ( _x3 , _y3 ), and the end point is ( _x4 , _y4 ), the x and y coordinates are expressed by the following equations, with the start point at t=0 and the end point at t=1.
x(t)=t ³ x ₄ +3t ² (1-t)x ₃ +3t(1-t) ² x ₂ +(1-t) ³ x ₁
y(t)=t ³ y ₄ +3t ² (1-t)y ₃ +3t(1-t) ² y ₂ +(1-t) ³ y ₁
When determining the area range, the area between the start point and the end point may be divided into about 10 parts and approximated by a plurality of straight lines connecting each point.

<Area Determination>
The area determination is a process of determining in which area, among areas set by constants, each object exists.
Since lanes are divided and set as areas, multiple areas overlap at points where lanes intersect, such as at forks and intersections. When multiple areas overlap, an area included in the route that the object has previously traveled is selected. When there are multiple routes, such as straight-ahead left-turn lanes, the object is considered to exist across multiple areas.
If an object suddenly appears within an intersection due to factors such as detection failure, it will exist at the intersection of multiple routes and areas, but in such cases it will also be treated as existing across multiple areas.

<Route determination>
Route determination is the process of determining which route the object is moving along when the area in which it exists has been determined. If multiple routes are possible, such as straight and left-turn lanes, all routes are treated as possible routes.
However, if it is determined that the vehicle cannot travel along the route as a result of the destination determination, it is assumed that there is no suitable route, and the destination is calculated based on the current speed, direction, and acceleration.

<Area location determination>
The area position determination is a determination of where in the area the object is located relative to the current position.
8A and 8B, the coordinates of the point on the area line where the object exists that has the smallest distance ΔD from the current position and the amount of deviation of the current position in the latitude and longitude directions (vertical and horizontal directions) (hereinafter referred to as the position offset) are calculated. The position offset is normalized so that the direction of the area line is due north (normalization makes the vertical direction 0).

The position of an object in an area is calculated using the distance from the upstream area, calculated using the coordinates of a point on the area. When restoring coordinates from the distance from the upstream area, a position offset is added to the coordinates calculated from the position on the area according to the direction of travel of the area on the coordinates.
Since calculating a normal from an arbitrary point to a cubic Bezier curve is a heavy load, a position offset may be calculated using a straight line that approximates the cubic Bezier curve.
As shown in FIG. 9, the position obtained by adding a position offset perpendicular to the direction of progress of the area (the direction of the tangent) is the unit position.

<Light color judgment>
The light color is determined by calculating the light color at the current date and time or the date and time of the future location using the time when the signal information was created and the current date and time. Note that the accuracy (error) of the signal information is not taken into consideration.
If there is a range for the number of seconds for each light color in the signal information (the shortest and longest do not match), it is basically treated as operating at the shortest time. However, if there is a light color for which the longest time is unknown, that light color will be treated as continuing to be displayed. Also, if a problem occurs with the integrity monitoring of the signal information, or if signal information is not received for three seconds (specified by a program constant), the light color received in the light color information will be treated as continuing to be displayed. If light color information cannot be received, all lights will be treated as being off.

<Behavior Judgment>
The current speed and acceleration are determined from the current and past speeds.
If the object's current speed is unknown, or if there is an object of the same group on the route, the speed of the nearest object is used (adjusted to the surrounding flow). If there is no object, the speed is the set speed (upper limit speed) of the existing area, and the direction is the direction of movement from the current position to the next area. However, if the situation is judged to require deceleration, the speed is reduced to a level at which deceleration is unnecessary.

If the current speed of an object is known, the object's speed and acceleration are calculated using a linear approximation formula based on the least squares method from the speed over the past three seconds (specified by a program constant). If any speed information within that three second period is unknown, an approximation is performed using only known information.

<Object collision detection>
A determination is made as to whether the future positions of all other objects whose routes may cross are in collision with the future positions of the specified object at all times calculated. For the specified object's future position at a specific time (future position T1 after t seconds), the determination is made for all future positions of other objects from "t-A" seconds after to "t+A" seconds after. For example, A is set to 10% of t (specified as a constant), and fractions are rounded up.

Based on the future position (current position) of the object at a target time (hereinafter referred to as T), the area in which the object may exist is calculated using the total length and total width according to the type of object.
Fig. 10 shows an example of the size of an object used in collision detection. Here, objects are classified into automobiles, buses, bicycles, two-wheeled vehicles, and pedestrians, and their overall length and width are determined in advance. The forward direction of an object (the direction of its overall length) is set to the traveling direction of the object at the relevant time. Internally, the area in which an object may exist is treated as a rectangle.

11 is a schematic diagram for explaining collision determination in detail. Using the future position of the specified object at time T and the future position just before time T (if the specified object does not exist, the future position just before time T), the polygon with the maximum area with each vertex is calculated, and similarly, the polygon with the maximum area is calculated for all other objects (other objects) with which the route may intersect, and it is determined whether the polygons overlap each other.
If there is even one overlap, it is considered to be a collision.

<Object avoidance behavior>
When the designated object performs an action to avoid the collision with another object, it is assumed that the designated object has performed an action to avoid the collision, and when the designated object does not perform the avoidance behavior, it is assumed that a collision will occur.

Figure 12 shows an example of the configuration of the road traffic control system shown in Figure 1 when communicating directly with vehicles using an IP network. In this example, the control module 11, traffic signal controller 12, hub 26, and mobile router 28 are housed in the housing 18. Signal information SA1 from the traffic signal controller 12, environmental information SA3 from the environmental sensor 14, and positioning information SA6 from the GNSS (Global Navigation Satellite System) system are input to the control module 11. In addition, an image of a vehicle taken by the vehicle light camera 21 is processed by the image processing unit 23-1, and vehicle information SA2a is input to the control module 11 via the hubs 25 and 26. Similarly, an image of a pedestrian taken by the pedestrian light camera 22 is processed by the image processing unit 23-2, and pedestrian information SA2b is input to the control module 11 via the hubs 25 and 26.

Driving information SA4 from autonomous vehicles 16-1, 16-2, 16-3, ... (here, autonomous vehicle 16-1 is taken as an example) is transmitted to base station 29 using a mobile line provided by a communications carrier, and is input from this base station 29 to control module 11 via LTE antenna 27, mobile router 28, and hub 26. Meanwhile, roadside information SA5 such as traffic light information SA1 and danger information SA5a is transmitted from control module 11 to base station 29 via hub 26, mobile router 28, LTE antenna 27, and mobile line, and is transmitted from this base station 29 to autonomous vehicle 16-1 with corresponding identification information SA4a.
According to this configuration, communication with autonomous vehicles 16-1, 16-2, 16-3, . . . can be performed using mobile lines provided by communication carriers such as mobile phone carriers.

Figure 13 shows an example of the configuration of the road traffic control system shown in Figure 1 when communicating with vehicles via a server using an IP network. In Figure 13, the same parts as in Figure 12 are given the same reference numerals, and detailed explanations are omitted. In this example, a distribution server 20A is provided on the cloud, and driving information SA4 from autonomous vehicles 16-1, 16-2, 16-3, ... (here, autonomous vehicle 16-1 is taken as an example) is transmitted to a base station 29 using a mobile line provided by a telecommunications carrier, and input from this base station 29 to the distribution server 20A. Then, from this distribution server 20A, via the base station 29, the driving information SA4 is input to the control module 11 via the LTE antenna 27, mobile router 28, and hub 26.

Meanwhile, roadside information SA5 such as traffic signal information SA1 and danger information SA5a is transmitted from the control module 11 to a base station 29 via the hub 26, the mobile router 28, the LTE antenna 27, and the mobile line, and is input from the base station 29 to a distribution server 20A. The roadside information SA5 is then transmitted from the distribution server 20A via the base station 29 to the autonomously driven vehicle 16-1 with the corresponding identification information SA4a.
According to this configuration, it is possible to communicate with autonomous vehicles via the distribution server 20A on the cloud using a mobile line provided by a communication carrier such as a mobile phone carrier. Also, danger information can be distributed from the distribution server 20A to autonomous vehicles traveling through multiple intersections.

Figure 14 shows a configuration example when a first (last) one mile management system is applied to the road traffic control system shown in Figure 1. In Figure 14, the same parts as in Figures 12 and 13 are given the same reference numerals, and detailed explanations are omitted. In this example, a system management server 20B is provided on the cloud, and driving information SA4 from autonomous vehicles 16-1, 16-2, 16-3, ... (here, autonomous vehicle 16-1 is taken as an example) is transmitted to a base station 29 using a mobile line provided by a telecommunications carrier, and is input from this base station 29 to the control module 11 via an LTE antenna 27, a mobile router 28, and a hub 26.

On the other hand, roadside information SA5, such as traffic signal information SA1 and danger information SA5a, is transmitted from the control module 11 to the base station 29 via the hub 26, mobile router 28, LTE antenna 27, and mobile line, and is then transmitted from the base station 29 to the autonomous vehicle 16-1 corresponding to the identification information SA4a.

In addition, the various types of detection information input to the control module 11 are transmitted to the base station 29 via the hub 26, mobile router 28, LTE antenna 27, and mobile line, and are input from the base station 29 to the system management server 20B.
According to this configuration, it is possible to communicate with autonomous vehicles using a mobile line provided by a communication carrier such as a mobile phone carrier. Moreover, the system management server 20B on the cloud can manage multiple roadside units via the base station 29, and can integrate and manage information on multiple intersections.

As described above, in the present invention, the control module recognizes vehicles around the intersection using input from the sensors (current position, vector), and predicts future positions from the current positions and vectors of vehicles around the intersection. It receives current position and route information from the autonomous vehicle, and calculates whether there will be contact with the autonomous vehicle based on the route of the autonomous vehicle and the predicted future positions of surrounding vehicles. If there is a possibility of contact, danger information is sent to the autonomous vehicle, so the accuracy of future predictions can be improved over a wide range that cannot be grasped by the autonomous vehicle alone. This can further improve traffic safety near intersections.

The circuit configurations and operation procedures described in the above embodiments are merely schematic illustrations to the extent that the present invention can be understood and implemented. Therefore, the present invention is not limited to the described embodiments, and can be modified in various forms without departing from the scope of the technical ideas set forth in the claims.

1...Roadside unit, 2...In-vehicle equipment, 11...Control module, 12...Traffic signal controller, 13...Multi-sensor module (mobile sensor), 14...Environmental sensor, 14a...Weather sensor, 14b...Road surface monitoring sensor, 15, 17...Wireless device, 16, 40, 41...Autonomous driving vehicle, 21...Vehicle light camera, 22...Pedestrian light camera, 23, 23-1, 23-2...Image processing unit, 31, 32...Signal pole, 3 3...vehicle, 34...right turn route, 35...right turn waiting point, 36...straight route, 37...intersection passing point, SA1...signal information, SA2...mobile object information, SA2a...vehicle information, SA2b...pedestrian information, SA3...environmental information, SA3a...weather information, SA3b...road surface information, SA4...driving information, SA4a...identification information, SA4b...location information, SA4c...route information, SA5...roadside information, SA5a...hazard information

Claims (8)

A system that predicts a future position of a moving object detected by a moving object sensor, determines a possibility of a collision with the moving object based on position information and route information received from an autonomous vehicle, and notifies the autonomous vehicle of the possibility of a collision,
a traffic light control device that predicts a color of a signal lamp when the autonomous vehicle approaches an intersection based on signal information from a traffic light control device, corrects a result of a collision possibility determination based on the predicted signal information and environmental information including road surface conditions from an environmental sensor, and classifies the collision risk into levels and notifies the user of the level of the collision risk ;
The determination of the possibility of collision is performed by forming a polygon having the maximum area with vertices at the future position of the designated object after a predetermined time and each point of the future position immediately before the future position, forming a polygon having the maximum area with respect to all other objects with which the traveling route may intersect, and determining that a collision will occur if the polygons overlap each other.
Road traffic control system. The road traffic control system according to claim 1, characterized in that the collision risk level is divided into a plurality of levels according to the deceleration required for the autonomous vehicle to avoid the risk. The road traffic control system according to claim 2, characterized in that the deceleration includes normal deceleration that allows the vehicle to stop before the predicted collision area, maximum deceleration that allows the vehicle to stop before the predicted collision area, deceleration due to the activation of an emergency brake that allows the vehicle to stop before the predicted collision area, and deceleration due to a damage mitigation brake that does not allow the vehicle to stop before the predicted collision area but performs maximum deceleration to mitigate damage. The road traffic control system according to claim 1, characterized in that the future position of the moving object is predicted from the current position and vector of the moving object detected by the moving object sensor, and the possibility of a collision with the moving object is determined from the position information and route information from the autonomous vehicle. The road traffic control system according to claim 1, characterized in that the traffic light information includes the light color of the traffic light and the time until the signal changes, and the traffic light information is used to predict the light color and the timing of the next signal change to control the autonomous vehicle, including maintaining its speed, decelerating, stopping, and starting, and brake operation and engine output suppression are performed according to the level of danger information that divides the risk of collision into multiple levels, thereby controlling deceleration or stopping to avoid collisions and contact. The road traffic control system according to claim 1, characterized in that the road surface conditions include wetness, snow cover, and ice on the road surface. The road traffic control system according to claim 1, characterized in that the mobile sensor is an imaging sensor or a LiDAR (light detection and ranging). The road traffic control system according to claim 1, characterized in that the moving object is a non-automated vehicle, a bicycle, or a pedestrian.