JP7394206B2 - Driving support device - Google Patents

Description

The present invention relates to a driving support device that understands the structure of roads and lanes based on information around the vehicle and map information, and controls the vehicle and outputs notifications to the driver.

Autonomous driving technology is being actively developed with the aim of eliminating traffic accidents, reducing the burden on the environment, and creating a secure, safe, and convenient mobility society that supports mobility for the elderly and others. Levels 1 to 5 defined by the Society of Automotive Engineers (SAE) are often used as levels of automatic driving. According to the SAE definition, vehicle control at level 3 or higher is the responsibility of the system. Therefore, in order to achieve level 3 or higher automated driving, it is necessary to control with more reliable information, and the use of maps is essential. High-precision maps prepared by prior surveying are generally used as maps. However, maps used for autonomous driving require information to be fresh, and maintaining that freshness costs a lot of money, so the range of high-precision maps that can be provided may be limited to expressways and major arterial roads. Therefore, highly reliable control is required on public roads even if high-precision maps are not available.

Therefore, for example, in Patent Document 1, an automatic driving control system is defined as "a driving support system that supports driving of a vehicle, which includes a position information acquisition unit that acquires position information of the vehicle, and a simple map that represents a simple map used for route guidance. a route generation unit that generates a travel route indicating which road the vehicle will travel on the simple map using the information and the position information; a surrounding map generation unit that uses position information to generate a surrounding map of the vehicle, which is a surrounding map that is more accurate than the simplified map, as surrounding map information while the vehicle is traveling; and the simplified map that includes the driving route. and a feature extraction unit that extracts road features from each of the road features and the surrounding map, and aligns the simple map and the surrounding map based on the road features, and calculates the position of the vehicle as a vehicle position. and a positioning unit that projects the driving route generated using the simple map information and the position information onto the surrounding map using the vehicle position and the driving route projected on the surrounding map. A path that generates a path for the vehicle to travel along the travel route based on the route, which is a travel trajectory and route of the vehicle in the surrounding map that is input to a vehicle control mechanism of an automatic driving vehicle. A driving support system equipped with a generating section is proposed.

Patent No. 6456562

In Patent Document 1, automatic driving that does not deviate from the road is realized using surrounding map information generated using sensor information and simple map information on a general road. However, the simple map information does not include connections between entry and exit lanes at intersections, and there is a problem in selecting the optimal driving lane to reach the destination. For example, if a driver turns right at an intersection and then turns left to reach a destination, when turning right at the first intersection, the driver's intention is to move to the far left lane of the road to which he or she is turning. . Therefore, it is necessary to calculate the connection relationship between entry and exit lanes at an intersection based on surrounding information from sensor information, and select the optimal driving lane to reach the destination.

The present invention has been made in view of the above points, and its purpose is to make it possible to select the optimal lane when planning a trip to a destination, and to make it possible to select the optimum lane according to the driver's intention on general roads including intersections. The objective is to obtain a driving support device that can realize autonomous driving.

The driving support device according to the present invention includes a sensor information acquisition unit that acquires surrounding information of the own vehicle detected by a sensor, and a lane information of a road based on information about at least one of a white line or a road edge included in the surrounding information. A sensor road/lane structure generation unit that generates a lane center point sequence and generates road and lane structure information based on the lane center point sequence, and a road and road lane on which the vehicle should drive based on the road and lane structure information. The vehicle is characterized by having a travel planning section that determines the travel planning section.

According to the present invention, it is possible to select an optimal lane based on a travel plan to a destination, and automatic driving according to the driver's intention can be realized on general roads including intersections. Further features related to the invention will become apparent from the description herein and the accompanying drawings. Further, problems, configurations, and effects other than those described above will be made clear by the following description of the embodiments.

FIG. 1 is a diagram illustrating the configuration of a vehicle equipped with a driving support device according to a first embodiment. FIG. 2 is a block diagram illustrating internal functions of the driving support device in the first embodiment. FIG. 3 is a block diagram illustrating internal functions of a sensor road/lane structure generation unit and a map synthesis unit shown in FIG. 2. FIG. 7 is a flowchart illustrating the processing content in the lane center point sequence generation unit. A diagram showing intersections and road IDs set for the intersections. The figure which shows an example of a simple map comprised by a map node and a map link. A diagram schematically showing a detection range of a sensor. A diagram showing a recommended travel trajectory when passing through an intersection, and entry and exit points of a lane-free section such as an intersection. FIG. 3 is a diagram illustrating a sequence of points indicating the boundary between an intersection and the outside of the road. A diagram illustrating information on white lines ahead of a right or left turn at an intersection. A diagram showing a trajectory of a vehicle turning right at an intersection. A diagram showing an example of an intersection. A diagram showing an example of an intersection. A diagram showing an example of an intersection. FIG. 3 is a diagram illustrating an example of generating a route by combining road/lane information with map information. FIG. 7 is a diagram illustrating another example of generating a route by combining road/lane information with map information. FIG. 7 is a diagram illustrating another example of generating a route by combining road/lane information with map information. FIG. 2 is a block diagram illustrating internal functions of a driving support device in a second embodiment. FIG. 3 is a diagram illustrating an example of a method for generating a lane center point. FIG. 3 is a diagram illustrating a method of calculating boundaries for lane increase/decrease. The figure which shows an example of the method of generating a virtual lane center point sequence from the approach point and exit point of an intersection. A diagram schematically showing the area around a toll gate on a toll road.

Hereinafter, embodiments of the present invention will be described in detail using the drawings. In addition, although the following embodiment explains the case where traffic is on the left side of the road as an example, it can also be applied to right side traffic.

<First embodiment>
FIG. 1 is a diagram illustrating the configuration of a vehicle equipped with a driving support device according to the present embodiment.
As illustrated in FIG. 1, the driving support device 3 is installed in a self-driving vehicle 1, and is driven by the cooperation between hardware such as a CPU and memory, and a software program executed by the hardware. Realized. The driving support device 3 cooperates with the map/locator unit U1, the sensor S, and the actuator D to form a driving support system.

The driving support device 3 obtains map information, route information, own vehicle position information of the own vehicle 1, and posture information of the own vehicle 1 from the map/locator unit U1. Then, surrounding information is obtained from the camera sensor S1 and the LiDAR sensor S2, which are the sensors S. Furthermore, own vehicle information is obtained from the vehicle sensor S3, and respective operation target amounts of the engine D1, which is the actuator D, the steering D2, the brake D3, etc. are determined.

The map/locator unit U1 includes a locator function 24 and a map transmission function 25. The locator function 24 receives GNSS (location information) and determines own vehicle position information and posture information, and the map transmission function 25 determines the own vehicle position information and attitude information. It includes a communication unit U2 that receives data 8. Note that the map data 8 may be a simple map for navigation (hereinafter referred to as a simple map) or a high-precision map having highly accurate information at the lane level.

FIG. 2 is a block diagram illustrating the internal functions of the driving support device.
The driving support device 3 inputs data from the map/locator unit U1 and the sensor S, and outputs the calculation results of the driving support device 3 to the speaker 111 and the actuator D. The driving support device 3 includes a map/route information acquisition unit 101, a position/orientation information acquisition unit 102, a sensor information acquisition unit 103, a sensor information integration unit 112, a self-position/orientation estimation unit 104, and a sensor road/lane structure. It includes a generation section 105, a map synthesis section 106, a route information addition section 107, and a control section 108. Further, the control unit 108 includes a travel planning unit 109 and a trajectory planning unit 110.

[Flow of data]
The map/route information acquisition unit 101 inputs data 2001 from the map/locator unit U1 and outputs data 2004 and 2005. The position and orientation information acquisition unit 102 inputs data 2002 from the map/locator unit U1 and outputs data 2016. The sensor information acquisition unit 103 inputs data 2003 from the sensor S and outputs data 2006 and 2017.

The sensor information integration unit 112 inputs data 2017 from the sensor information acquisition unit 103 and data 2021 from the self-position and orientation estimation unit 104, and outputs data 2019 and 2020. The sensor road/lane structure generation unit 105 inputs the data 2019 from the sensor information integration unit 112 and the data 2007 from the self-position and orientation estimation unit 104, and outputs the data 2009.

The self-position and orientation estimation unit 104 inputs data 2016 from the position and orientation information acquisition unit 102, data 2006 from the sensor information acquisition unit 103, and data 2020 from the sensor information integration unit 112, and inputs data 2007, 2008, 2018. Output. The map synthesis unit 106 inputs data 2005 from the map/route information acquisition unit 101, data 2018 from the self-position/orientation estimation unit 104, and data 2009 from the sensor road/lane structure generation unit 105, and generates data 2010, 2012 is output. The route information addition unit 107 inputs data 2004 from the map/route information acquisition unit 101, data 2008 from the self-position and orientation estimation unit 104, and data 2010 from the map synthesis unit 106, and outputs data 2011.

The control unit 108 inputs data 2011 from the route information addition unit 107 and data 2012 from the map synthesis unit 106, and outputs data 2014 and 2015. The travel planning unit 109 receives data 2011 from the route information addition unit 107 and data 2012 from the map synthesis unit 106, and outputs data 2013 and 2014. The trajectory planning section 110 inputs data 2011 from the route information addition section 107, data 2012 from the map synthesis section 106, and data 2013 from the travel planning section 109, and outputs data 2015. The speaker 111 constitutes a notification means that inputs data 2014 from the trip planning section 109 and outputs a notification to the driver. Actuator D receives data 2015 from trajectory planning section 110 and controls the vehicle.

[Explanation of functional blocks and data contents]
The data 2001 from the map/locator unit U1 is assumed to be a simple map and route information, and is acquired by the map/route information acquisition unit 101, converted as necessary, and stored in a memory. FIG. 6 is a diagram showing an example of a simple map composed of map nodes and map links. As shown in FIG. 6, the simple map has a map node F001 and a map link F002.

A node ID is set in the map node F001, and latitude and longitude are stored as position information of the node. Further, altitude information is stored in the map node F001. Furthermore, information such as the link angle F003 between map links and the number of links is stored in the map node F001. In the example of FIG. 6, the number of links stored in map node F001 is four. There are also four link angles F003 stored in the map node F001.

Further, the map node F001 stores the road type, and in the case of an intersection, information such as a crossroad, a T-junction, a six-way intersection, etc. is stored as the intersection type. In addition, the map link F002 stores the connection relationship of IDs between map nodes, the number of lanes in which own vehicle 1 is traveling, and the number of lanes in the opposite lane, and also stores the speed limit value, the position on the map link, and the map Link shape information etc. are stored.

The route information represents a list of IDs of the map node F001, and represents information on which map node to go through to reach the destination. Alternatively, in addition to IDs, pairs of latitude and longitude may be listed as location information of map nodes. Further, in addition to the position information of the map node, guidance information about the road ahead or the direction of right or left turn, which is notified for navigation, may be acquired.

Examples of conversion by the map/route information acquisition unit 101 include converting information in a global coordinate system such as latitude and longitude to a plane coordinate system. This has the advantage that the distance calculation in the latter stage can be performed using the Pythagorean theorem, which is easier than the Hubeny formula, which is commonly used for distance calculation in the global coordinate system. Note that the simple map may be obtained from V2X.

Data 2002 from the map/locator unit U1 is information from a locator, and assumes latitude and longitude as position information, and assumes orientation as attitude information. The position and orientation information acquisition unit 102 acquires the data 2002, performs conversion as necessary, and stores the converted data in memory. Examples of conversion include conversion to a plane coordinate system, similar to the map/route information acquisition unit 101.

Data 2003 from sensor S is composed of data from camera sensor S1, LiDAR sensor S2, and vehicle sensor S3 in FIG. Data from camera sensor S1 includes position information, size information, and type information of three-dimensional objects, position information of white lines, broken line and color type information, roadside position information, position information and color information of traffic lights, and stop line information. , location information of crosswalks, position information and type information of signboards, etc.

FIGS. 5 and 14 are diagrams showing an example of an intersection. FIG. 5 is a diagram showing intersection lanes and road IDs set at the intersection. FIG. 14 is a diagram showing an example of an intersection with traffic lights. .
An example of the white line is F013 in FIG. 5. The white line may be a point sequence obtained by extracting a sequence of consecutive end points from a point group detected by edge detection, or may be in the form of approximate parameters. An example of a roadside is F010 in FIG. An example of the stop line is F014 in FIG. 5. An example of a crosswalk is F011 in FIG. 5. Examples of signage include speed limit signs, stop lines, lane reductions, etc. An example of a traffic light is F021 in FIG. 14.

In addition, the data from the LiDAR sensor S2 is assumed to include position information and size information of three-dimensional objects, type information such as white line position information and broken lines, road edge position information, stop line position information, crosswalk position information, etc. do. Further, the data from the vehicle sensor S3 is assumed to be own vehicle speed information, steering angle information, yaw rate information, acceleration information, etc. Note that the sensor S may be any sensor as long as it can obtain data necessary for understanding the surrounding environment of the vehicle 1 similar to the above, and may be a radar, C2X, V2X, or the like.

A sensor information acquisition unit 103 acquires information from each of these sensors, converts it as necessary, and stores it in a memory. As an example of conversion, when the unit systems and resolutions of data from multiple sensors are different, the data is converted into a unified format so that it can be easily handled in subsequent processing.

The data 2017 from the sensor information acquisition unit 103 is assumed to be sensor information in a format with a unified data unit system and resolution, and the sensor information integration unit 112 integrates the data so that it is easier to handle in subsequent processing. As an example of integration, if the data coordinate systems and origins of the coordinate systems are different for each sensor, they are unified into the same coordinate system and origin for ease of handling in subsequent processing.

In this embodiment, it is assumed that the coordinate system relative to the own vehicle is used and will be described below. In addition, when the data transmission cycles of the plurality of sensors are different from each other, they are synchronized with the processing cycle of the driving support device 3 so that it can be easily treated as a unified time in subsequent processing. Synchronization is calculated by estimating the amount of movement based on vehicle speed information and steering angle. Then, if each sensor detects the same three-dimensional object, white line, roadside, etc. in the synchronized information, the same ID is set as the same information, the data is integrated, and output.

In addition, based on the own vehicle position and attitude information, which is data 2021 from the own position and attitude estimation unit 104, past sensor data 2017 is recorded in the memory in the form of global coordinates, and interpolation processing is performed when data is missing, and An integration process based on empirical rules may also be implemented. For example, in the case of white line information, a method can be considered in which past white line position information and newly detected white line position information are compared, it is determined whether they are the same white line, and the ID is tracked. As a result, even in a situation where little sensor information is detected, the accumulated information can be utilized, and ID identification accuracy can be improved and integrated processing can be performed. Furthermore, unlike data such as camera sensor S1 and LiDAR sensor S2 that are based on the own vehicle relative coordinate system, it is known that the information from C2X and V2X is data that is based on the global coordinate system. , based on the own vehicle position and orientation of the data 2021, is converted to the own vehicle relative coordinate system, and is integrated with the data of the camera sensor S1 and LiDAR sensor S2.

The data 2019 from the sensor information integration unit 112 includes position information of three-dimensional objects, white line position information, roadside position information, etc., which have undergone synchronization of processing cycles, data integration, and unification of coordinate systems. - The lane structure generation unit 105 understands the boundaries between roads and lanes, and calculates connection relationships between roads and lanes. As specific examples of road and lane boundary areas, white line information F013, crosswalk information F011, stop line information F014, and roadside information F010 that can be detected by sensors in a situation where the own vehicle 1 is present as shown in FIG. can be mentioned.

Note that, as shown in FIG. 5, a road represents an area divided by lane changes, intersections, branches, etc., such as areas R01 to R06, and is managed by ID. A lane represents information in units of L01 to L13 included in regions R01 to R06, and is managed by ID.

The connections between these roads and lanes refer to the connections between the regions R01 to R06 before and after, and the connections between the lanes L01 to L13 between the front and back, and the left and right sides of the lanes. For example, based on the position of the own vehicle, the front of region R01 is connected to region R02, and the rear of region R02 is connected to region R01. Further, a lane L03 is connected to the right side of the lane L02, and a lane L02 is connected to the left side of the lane L03. Furthermore, lane L06 and lane L07 are connected to the front of lane L02, and lane L02 is connected to the rear of lane L06 and lane L07.

Note that the sensor road/lane structure generation unit 105 generates an output based on the information detected within the detection range F015 of the sensor S as shown in FIG. Regenerate information.

FIG. 7 is a diagram showing the state before and after the host vehicle 1 enters the right turn lane.
When the vehicle 1 moves from in front of the right-turn lane at the intersection to above the right-turn lane at the intersection in FIG. 7, the detection range changes from F015 to F020, making it possible to obtain more sensor information within the intersection.

The sensor road/lane structure generation unit 105 manages the data 2017 from the sensor information acquisition unit 103 in association with the structure of roads and lanes, and additionally generates additional information that can be calculated only from the sensor information.

As an example of management in which the data 2017 is linked to the structure of roads and lanes, it is conceivable to associate the white line information of the data 2017 with each lane, and to associate the roadside information of the data 2017 with the road. Examples of additional information include the center point sequence of the lane, the radius of curvature, the speed limit, the length of the lane, and the distance until the lane disappears or appears. As an example of the distance until the lane disappears, the distance until the lane L02 disappears in Figure 5 is the sum of the length of lane L01 and the length of lane L02 when the own vehicle is at the starting position of lane L01. This is the result.

Furthermore, as shown in the detection range F015 in FIG. 7, the data 2017 is generally known that the farther the distance from the boundary area of the detection range or the own vehicle 1, the larger the detection error. Additionally, false detection may occur depending on the driving conditions. Therefore, it is preferable to add reliability information to the road/lane structure information 2009 in consideration of the characteristics of these sensors S. Further, the generation accuracy of road and lane structures may be improved by utilizing the own vehicle position and attitude, which is the data 2007 from the own position and attitude estimating unit 104, and using past information in the same way as the sensor information integrating unit 112.

The data 2016 from the position and orientation information acquisition unit 102 assumes latitude and longitude as position information and the orientation as attitude information, and the data 2006 from the sensor information acquisition unit 103 assumes vehicle speed information, steering angle, etc. Assuming the information, the self-position and orientation estimation unit 104 estimates the latitude and longitude as the final positional information of the own vehicle with increased accuracy for the driving support device 3, and estimates the direction as the final positional information of the own vehicle. do.

The data 2016 based on the information from the locator is longer than the processing cycle of the driving support device 3, such as a transmission period of 100 milliseconds to 1 second. Correction of position and posture is required. The own vehicle position and attitude are corrected by calculating the amount of movement using the own vehicle speed information, steering angle information, yaw rate information, etc. of the data 2006 from the sensor information acquisition unit 103.

In addition, the accuracy of the own vehicle position and posture information within the lane may be increased using the white line position information of the data 2020 from the sensor information integration unit 112. Furthermore, the accuracy of the own vehicle position and attitude information on the road may be increased using the roadside information of the data 2020. If the accuracy of the own vehicle's position increases, if the map synthesized by the map synthesis unit 106 is highly accurate, the error when converting to the own vehicle relative coordinate system can be minimized, and more accurate information can be transmitted to the control unit 108. There are advantages to using it.

The data 2009 from the sensor road/lane structure generation unit 105 assumes the road and lane structure information described above, and the data 2018 from the self-position and orientation estimation unit 104 assumes the own vehicle position corrected for the driving support device 3. Assuming the posture, the data 2005 from the map/route information acquisition unit 101 is assumed to be map information, and the map synthesis unit 106 calculates final road and lane structure information with high reliability that is handled by the control unit 108.

The map synthesis unit 106 extracts map information of a necessary area from the data 2005 based on the own vehicle position and orientation of the data 2018, and determines the consistency of the information with the data 2005 based on the road and lane structure information of the data 2009. Verify. If there is a discrepancy between the pieces of information, the road and lane structure information added to the data 2009 with low reliability information uses the information in the data 2005 to increase the reliability. Thereby, more reliable information can be provided to the control unit 108, and control that meets the driver's intention becomes possible. Note that if map information is not provided from the map/locator unit U1, the map synthesis unit 106 outputs the road and lane structure information from the sensor road/lane structure generation unit 105 as data 2010, 2012, thereby improving the driving performance. Support can be continued.

The data 2004 from the map/route information acquisition unit 101 assumes route information, and the data 2008 from the own position and orientation estimation unit 104 assumes the own vehicle position and orientation corrected for the driving support device 3, The data 2010 from the synthesis unit 106 assumes road and lane structure information with increased reliability. The route information of the data 2004 is assumed to be a map-independent route so that the data 2010 from the map synthesis unit 106 can be matched with the coordinate-related information of the road and lane structure information expressed in vehicle-relative coordinates. do.

As an example of a route that does not depend on a map, a list of latitudes and longitudes is used as the route, and the latitude and longitude of the route are converted to vehicle-relative coordinates based on the vehicle position and orientation of data 2008 from the self-position and orientation estimation unit. do. Using the converted route coordinates and the lane center point information linked to the lane structure information of the data 2010, it is determined which road the route belongs to.

As a result, the route information addition unit 107 outputs an ID list of road and lane structure information in the data 2010. Further, as the guidance information from the map/route information acquisition unit 101, the road ahead and right/left turn guidance may be added to the data 2011 of the route information addition unit 107.
This has the advantage that the guidance information can be utilized even in areas that cannot be detected by the sensor, and can be used as a basis for determining which lane the travel planning section 109 of the control section 108 should move to.

The data 2011 from the route information adding unit 107 assumes the ID list of road and lane structure information corresponding to the data 2010 and guidance information from the navigation, and the data 2012 from the map synthesis unit 106 assumes roads with increased reliability. and lane structure information. The travel planning unit 109 of the control unit 108 determines the road on which the vehicle should travel based on the route information in the data 2011, and determines the lane on which the vehicle should travel on the basis of the lane structure information in the data 2010. Alternatively, the lane in which the vehicle should travel is determined based on the route information in the data 2011. The effect of the driving support device 3 in this embodiment can be increased by selecting the optimal lane in consideration of which lane to drive in when turning right or left at an intersection in order to reach the destination of the route. Can be maximized.

The data 2013 from the travel planning unit 109 is assumed to be a list of lane IDs to drive in, and the data 2012 from the map synthesis unit 106 is assumed to be the speed limit, radius of curvature, and lane center point linked to lane structure information. The trajectory planning unit 110 determines the trajectory that the vehicle should travel several seconds ahead and the target speed of the vehicle.

The data 2014 from the travel planning unit 109 assumes the lane ID and lane structure information in which the vehicle should travel, and the speaker 111 notifies the driver of which lane the vehicle should travel in as a warning. The data 2015 from the trajectory planning unit 110 assumes the trajectory that the vehicle should travel and the target speed of the vehicle, and the actuator D determines and controls the target steering angle, target brake fluid pressure, and target engine torque. For the purpose of realizing an automatic driving function as a driving support system, it is preferable to output data to the speaker 111 and the actuator D and perform control.
Further, if the purpose is to realize safety support as a driving support system, it is preferable to output data only to the speaker 111 to alert the driver.

[composition]
FIG. 3 is a block diagram showing the internal functions of the sensor road/lane structure generation section 105 and the map synthesis section 106.
In FIG. 3, data from the sensor information integration unit 112 and data from the self-position and orientation estimation unit 104 are input to the sensor road/lane structure generation unit 105, and the calculation results of the sensor road/lane structure generation unit 105 are used as the input to the map synthesis unit. 106. The data from the map/route information acquisition unit 101, the data from the self-position/orientation estimation unit 104, and the data from the sensor road/lane structure generation unit 105 are input to the map synthesis unit 106. 106 calculation results are output.

The sensor road/lane structure generation unit 105 includes a lane center point sequence generation unit 302, a region division determination unit 303, a lane information generation unit 304, a road information generation unit 305, and a sensor additional information generation unit 306. Furthermore, the map synthesis section 106 includes a map selection section 307, an error correction section 308, a lane information generation section 309, a road information generation section 310, a simple lane center point sequence generation section 311, and a map additional information synthesis section 312.

[Data flow]
In FIG. 3, data 2019 from the sensor information integration unit 112 and data 2007 from the self-position and orientation estimation unit 104 are input to the lane center point sequence generation unit 302, and data 4005, 4006, and 4007 are output.

Data 4005 from the lane center point sequence generation unit 302 is input to the area division determination unit 303, and data 4008 and 4009 are output. Data 4006 from the lane center point sequence generation unit 302 is input to the lane information generation unit 304, and data 4010 is output. Data 4009 from the area division determination unit 303 and data 4010 from the lane information generation unit 304 are input to the road information generation unit 305, and data 4011 is output. Data 4011 from the road information generation section 305 is input to the sensor additional information generation section 306, and data 2009 is output. Data 2005 from the map/route information acquisition unit 101 and data 2018 from the self-position and orientation estimation unit 104 are input to the map selection unit 307, and data 4015, 4023, and 4024 are output.

Data 2009 from the sensor additional information generation section 306 and data 4015 from the map selection section 307 are input to the error correction section 308, and data 4016 is output. Data 4024 from the map selection section 307 and data 4016 from the error correction section 308 are input to the lane information generation section 309, and data 4017, 4018, and 4019 are output. Data 4017 from the lane information generation section 309 is input to the road information generation section 310, and data 4022 is output.

Data 4023 from the map selection unit 307 and data 4019 from the lane information generation unit 309 are input to the simple lane center point sequence generation unit 311, and data 4020 is output. Data 4018 from the lane information generation section 309, data 4022 from the road information generation section 310, and data 4020 from the simple lane center point sequence generation section 311 are input to the map additional information synthesis section 312, and data 4021 is output.

[Explanation of functional blocks and data contents]
The data 2019 from the sensor information integration unit 112 is assumed to be white line position information, roadside position information, etc. that have undergone processing cycle synchronization, data integration, and coordinate system unification, and is sent to the lane center point sequence generation unit 302. The positions of the center point arrays of multiple lanes located within the sensor detection range are calculated. Note that the data 2007 from the own position and orientation estimation unit 104 assumes the own vehicle position and orientation, and utilizes the history of the lane center point sequence generated by the lane center point sequence generation unit 302 to determine the lane center point to be output. Try to stabilize the line. Lanes include all lanes located within the sensor detection range, for example, in addition to the lane in which the own vehicle is traveling, lanes of oncoming traffic, lanes of adjacent lanes running parallel to each other, and lanes of intersecting roads. etc. are included. The lane center point is the lane center point of these lanes, that is, the center point in the lane width direction for each lane, and the lane center point column is the lane center point of each lane in the direction in which the lane extends. It is a series of connected points.

The lane center point sequence generated by the lane center point sequence generation unit 302 using the own vehicle position and orientation is converted into a global coordinate system and held within the lane center point sequence generation unit 302. Then, in the next processing cycle of the driving support device 3, the position of the lane center point sequence in the history is converted from the own vehicle position and attitude at that time to the own vehicle relative coordinate system, and the lane center point sequence calculated at the current time is converted to the own vehicle relative coordinate system. Integrate. As a result, even if there is a processing period in which the lane center point sequence is momentarily interrupted, the lane center point sequence can be continuously provided to the control unit 108, making it possible to stabilize the entire driving support system.

Figure 8 is a diagram showing the recommended travel trajectory when passing through an intersection, and the entry and exit points of lane-free sections such as intersections. Figure 9 shows the road at the intersection before and after entering the right turn lane. FIG. 10 is a diagram showing a dot sequence indicating the boundary with the outside, and is a diagram showing a dot sequence indicating the position of the white line before and after entering the right turn lane.
The lane center point sequence generation unit 302 of this embodiment assumes that the recommended travel trajectory within an intersection is also a virtual lane and generates a lane center point sequence, as shown in F016 of FIG. An example of the road edge position information at an intersection is the position of a point sequence indicating the boundary between the intersection and the outside of the road, such as F022 in FIG. 9 . Further, as an example of the position information of the white line at the intersection, information on the white line at the destination after turning right or left at the intersection is included, such as F025 in FIG. 10.

Data 4005 from the lane center point sequence generation unit 302 assumes a center point sequence of a plurality of lanes, and the area division determination unit 303 determines lane increase/decrease boundaries based on the positional relationship of each lane center point sequence. As an example of lane increase, the boundary is defined as the area R01 and R02 in FIG. 5. Calculation of boundaries for lane increase/decrease will be explained using FIG. 20.

FIG. 20 is a diagram illustrating a method for calculating boundaries between lane increases and decreases.
There are lane center point rows F206 and F207 in FIG. 20, and perpendicular lines are drawn from the respective start point F203 and end point F204 to each lane center point row. Next, it is determined whether the distances F201 and F202 from the point of intersection with the perpendicular line exceed a certain threshold value, and if the distances F201 and F202 from the point of intersection with the perpendicular line exceed a certain threshold value, it is determined as a boundary for lane increase/decrease. In the example of FIG. 20, if only section F201 exceeds a certain threshold, F205 becomes the boundary for lane increase/decrease.

Furthermore, the data 4005 from the lane center point sequence generation unit 302 assumes an entry point F022 of a lane-free section such as an intersection shown in FIG. Based on the exit point and exit point, the boundary between a laneless section with no lanes and a lane section with other lanes is determined.

As examples of boundaries between lane-free sections and other boundaries, the boundaries between areas R02 and R04, between areas R04 and R03, between areas R04 and R05, and between areas R04 and R06 shown in FIG. Boundary. Based on the boundaries calculated by the area division determination unit 303, the lane information generation unit 304 and road information generation unit 305 in the subsequent stages assign IDs to the road and lane structure information.

Alternatively, the lane-free section including the intersection section and other boundaries may be determined in a composite manner from the position information of the signal from the sensor information integration unit 112, the stop line, and the position of the crosswalk.

In addition to intersections, lane-free sections also include entrances and exits of toll gates on toll roads. FIG. 22 is a diagram schematically showing the area around a toll gate of a toll road. In the case of a toll plaza, the position information and size information of the toll plaza gate F046 are acquired from the sensor information integration unit 112, and the upstream entry point F041 and exit point F042 of the toll plaza gate F046, and the downstream entry point F044 are obtained. It is preferable to calculate the exit point F045. By dividing the boundaries in this way, it is possible to decide how to change the driving plan at points where the lane to be selected changes, such as increase or decrease of lanes or intersections, which has the effect of making it easier to handle in later stage logic.

Returning to the explanation of FIG. 3.
Data 4008 from the area division determination unit 303 assumes lane increase/decrease boundary information, lane-free section and other boundary information, and data 4006 from the lane center point sequence generation unit 302 assumes position information of the lane center point sequence. Based on this assumption, the lane information generation unit 304 generates lane information. As shown in Figure 5, lane information is defined as lanes with white lines such as L01 to L03, L04, L05, and L11 to L13, and lanes without white lines such as inside intersections such as L06 to L10. The section is defined as a virtual lane.

Lane information refers to information that assigns lane IDs such as L01 to L13 and indicates the connection relationship between the front and rear and left and right sides of each lane. Furthermore, one lane is attached with information related to the lane, such as the lane center point array position, white line information, and signal information. The lane includes a plurality of lanes L01 to L13. Each of these lanes is attached with information regarding the structure of the lane, such as information indicating the front/back and left/right connections between lane sections, and the position of lane center point rows. ing.

The lane information generation unit 304 first generates a lane ID based on the position information of the lane center point sequence and the boundary information from the area division determination unit 303. Then, left and right connection information is generated based on the lateral positional relationship of the center point rows of each lane. Further, front and rear connection information is generated based on the positional relationship in the traveling direction of the center point array of each lane. Finally, the lane center point row position and white line information are linked to each lane. The lateral positional relationship of each lane center point row represents, for example, the order in which lanes are arranged from the left, and thus indicates which lanes are arranged in which order on the road.

Data 4009 from the area division determination unit 303 assumes lane increase/decrease boundary information, lane-free section and other boundary information, and data 4010 from the lane information generation unit 304 assumes lane ID, front/rear/left/right connection relationships. Based on this assumption, the road information generation unit 305 generates road information using these data 4009 and 4010. The road information is defined as regions R01 to R06, as shown in FIG.

The road information is assigned a road ID such as R01 to R06, and refers to information representing the connection relationship before and after each road. Furthermore, a plurality of lane IDs associated with one road are managed. Furthermore, each road has roadside position information associated with it.

The road information generation unit 305 first generates a road ID based on the boundary information from the area division determination unit 303. Then, connection information before and after each road is generated. Furthermore, a lane ID associated with one road is calculated from the data 4010 and associated. Finally, the roadside position information associated with one road is linked.

Data 4011 from the road information generation unit 305 assumes road information and lane information linked thereto, assumes data 4007 from the lane center point sequence generation unit 302, and the sensor additional information generation unit 306 generates these data 4011. and 4007 to generate additional information. Additional information is generated based on the lane center point sequence, and data 4011 is used to comprehensively determine which road ID is to be linked to which lane ID.

An example of the additional information includes calculating the radius of curvature based on the position of the lane center point sequence in the data 4007 from the lane center point sequence generation unit 302. As an example of how to determine the radius of curvature, three representative points are selected from the lane center point sequence, and a circle passing through those points is calculated.

In addition, the distance until the lane information ID is switched is calculated based on the position of the lane center point sequence, and the position of signal information, stop line, and crosswalk information from the sensor information integration unit 112 is converted into road information and lane information. Examples include linking. Further, considering the characteristics of the sensor S, reliability information is linked to road information and lane information according to the detection edge and detection distance of the detection range F015 as shown in FIG.

The data 2018 from the self-position and orientation estimation unit 104 assumes the own vehicle position, the data 2005 from the map/route information acquisition unit 101 assumes map information, and the map selection unit 307 assumes the own vehicle position of the data 2018. Based on this, a map around the own vehicle position is selected from the data 2005. The selected map information includes the position information of the map node F001 shown in FIG. 6, the link angle F003 between the map links, the number of links, the road type, the connection relationship of the ID between the intersection type and the map node of the map link F002, and the traveling lane. Assuming the number of lanes and the number of opposing lanes.

The data 2009 from the sensor additional information generation unit 306 assumes road and lane structure information with additional information, and the data 4015 from the map selection unit 307 assumes map information selected based on the own vehicle position. The correction unit 308 compares these two pieces of information and corrects the road and lane structure information in the data 2009 that does not match the map information. Since the road and lane structure information is based on a sensor whose accuracy decreases as the distance from the sensor detection range or the distance from the own vehicle increases, the reliability of the road and lane structure information changes depending on the obtained position.

On the other hand, map information has issues with its freshness, and the shape of the road may differ from the actual road due to construction work. When reliability can be obtained from map information, the reliability of road and lane structure information based on sensor information is low, and when reliability from map information is high, there is a discrepancy between map information and road and lane structure information. For example, correct road and lane structure information. As a result, it is possible to reduce the possibility that only map information that has a problem with freshness will be trusted and the final result will be incorrect.

The data 4016 from the error correction unit 308 assumes corrected road and lane structure information based on map information, and the data 4024 from the map selection unit 307 assumes the number of traveling lanes, road types, intersection types, and number of links. Based on this assumption, the lane information generation unit 309 generates lane information that is not included in the data 4016. Then, the connection relationship between the generated lane information and the lane information of data 4016 is calculated. Finally, the newly generated lane information and connection information are added to the data 4016, and data 4017 and 4019 are output.

Data 4017 from the lane information generation unit 309 is synthesized in the road information generation unit 310, assuming road and lane structure information having lane information that is a combination of lane information based on sensor information and lane information based on map information. The road information to which the lane information belongs is generated based on the map information. Then, the connection relationship between the generated road information and the road information of data 4017 is calculated. Finally, the newly generated road information and connection information are added to the data 4017, and data 4022 is output.

Data 4019 from the lane information generation unit 309 assumes lane information based on sensor information and lane information based on map information, and data 4023 from the map selection unit 307 assumes position information of map nodes and link angles between map links. , a simple lane center point sequence generation unit 311 generates a simple lane center point by utilizing the number of lanes of travel lanes.

FIG. 19 is a diagram illustrating an example of a method for generating a lane center point.
For example, in the scene where the vehicle F100 turns right at an intersection shown in FIG. 19, the lane center point is generated from the map node F109 at a distance F105, which is half the size of the intersection, that is, half the size of the intersection. A perpendicular line F106 is drawn down to the link F107, and an exit point F110 is set at a point at an estimated road center distance F104 from the map link F107.

Note that the estimated road center distance F104 is assumed to be (number of traveling lanes x 3.5 m)/2 of the map link F107. The reason for setting the road center is to set the exit point as a guide, rather than generating the exit point at the lane level, since the simple map itself is not information whose accuracy is guaranteed. Further, the distance F105 of intersection size/2 is set as the intersection size, which is the sum of the lane widths of each lane of the lane in which the host vehicle F100 is present. Note that the link angle F103 of F101 and F107 is used.

Then, the intersection angle at the point F113 where the extension line F112 of the road to which the exit point F110 included in the data 4019 belongs and the extension line F111 of the lane to which the entry point F102 belongs is defined as the link angle F103, and the intersection angle included in the data 4019 is set as the link angle F103. An angle F103 between the approach point F102, the exit point F110, and the point F113 is approximated by a circular arc, a clothoid curve, a spline curve, a Bezier curve, etc., and a lane center point is generated. Note that this information does not control the process until exiting the intersection based on the generated lane center point, but is used as a guide for entering the intersection, and after entering the intersection, the own vehicle F100 uses the white line information and road edge information on the exit side of the intersection. This will be used as a link until detection. Note that although the intersection exit point is one point, the intersection entry points are assumed to be multiple points, and a simple lane center point is generated for each lane entering the intersection.

The data 4022 from the road information generation unit 310 is assumed to be road information obtained by combining the road information of the data 2009 with the data 4015, and the data 4018 is assumed to be lane information obtained by combining the lane information of the data 2009 with the data 4015. The additional information synthesis unit 312 uses these data to generate additional information that is not included in the road and lane structure information of the sensor road/lane structure generation unit 105 and exists only in the map, that is, additional information that exists only in the map. This additional information is combined with the road and lane structure information from the sensor road and lane structure generation unit 105. As an example of additional information that only exists on a map, consider the road type and speed limit on the map. Furthermore, the data 4020 from the simple lane center point sequence generation section 311 is also synthesized with the road and lane structure information in the map additional information synthesis section 312. Then, data 4021 is output as the final output of the map synthesis section.

[Explanation of Figure 4]
A detailed example of the lane center point sequence generation unit 302 will be described using the flowchart of FIG. 4.
First, based on the white line information F024 shown in FIG. 10, center point sequence candidates are generated for all pairs of detected white line point sequences through the processes S01 to S04. This process is for sections where white lines can be detected. The white line dot sequence here represents a series of continuously connected white lines, and two of the white lines are made into a pair. An exception will be made in the case of broken lines, and if the line continues continuously, it will be treated as one of the white lines.

Next, in steps S05 to S08, a virtual lane center point is calculated based on the white line information F024 shown in FIG. 10 and the road edge information F022 shown in FIG. 9 as processing for a laneless section where no white line can be detected. Finally, the lane center points from S01 to S04 and the virtual lane center points from S05 to S08 are integrated in S09. Thereby, lane structure information indicating the connection relationship between lanes with a lane-free section in between can be generated.

The processing of S01 to S04 in FIG. 4 will be explained.
S01 is a repetition condition for determining whether all pairs of white line point sequences have been processed. In a section where a white line can be detected, one lane is composed of two white line dot arrays, so in this process, the pair is repeatedly processed.

Next, in S02, pairs of white line dot columns are selected. As in the example of FIG. 10, not only the rows of white line dots located continuously in the direction of travel of the host vehicle, but also the rows of white line dots on the right/left turn side after exiting an intersection located perpendicularly are subject to processing.

In S03, lane center point sequence candidates are generated from the selected pair of white line point sequences.

Next, in S04, only necessary center points are narrowed down from the lane center point candidates. For example, whether the distance between the white line dots is close to the general lane width of 3.5 m is used as an index, and if it is extremely short or long, it is excluded from the subsequent processing. This makes it possible to reduce the processing time at the subsequent stage. Also, based on the road edge information, it is determined whether the position of the lane center point array candidate is outside the road edge when viewed from the own vehicle, and the lane center point array candidate is excluded from the lane center point array candidates.

Next, the processing of S05 to S08 will be explained.

S05 is a condition for detecting a lane-free section. For example, a pattern in which the white line information F024 in FIG. 10 is interrupted at the intersection entrance, and a white line appears at the intersection exit exit on the straight-ahead or right-left turn side is determined and determined to be a lane-free section. In addition, the shape pattern of the intersection may be determined using the road edge information F02 2 in Figure 9 to determine a lane-free section, or the lane-free section may be determined from a combination of white line information, a stop line in front of the intersection, and a crosswalk. It's okay. Further, in order to increase the accuracy of the judgment, information on the sign indicating the right/left turn method F018 may be used as a feature point of the intersection, as shown in FIG. 12. Furthermore, in addition to the stop line before the intersection as shown in FIG. 13, the information on the sign of the stop line F019 inside the intersection may be used, or the signal information F021 as shown in FIG. 14 may be used. FIG. 12 is a diagram showing the right/left turn method F018 marked inside the intersection, and FIG. 13 is a diagram showing the stop line F019 inside the intersection.

In S06, the intersection entry point is calculated. The entry point refers to, for example, F022 shown in FIG. 8, and selects the end point of the lane center point of S01 to S04 calculated from the white line F024 in FIG. 10.

In S07, the exit point of the intersection is calculated. The exit point refers to F017 shown in FIG. 8, and similarly to S06, the end point of the lane center point of S01 to S04 calculated from the white line F024 of FIG. 10 is selected. In addition, if the white line F024 in Figure 10 cannot be detected but the road edge information F022 in Figure 9 can be detected, the interval F023 between the road edge points is assumed to be the road width, the number of lanes is estimated from the width, and the exit point is generated. It's good to do that. Further, based on the position F021 of the signal information in FIG. 14, only the exit point of the leftmost lane after a right turn may be estimated. It is best to consider these factors in combination and calculate a stable output.

In S08, a combination of the approach point calculated in S06 and the exit point calculated in S07 is determined, and a virtual lane center point sequence is calculated. The virtual lane center point sequence refers to, for example, the center point sequence indicated by dotted lines within the intersection in FIG. 8. A virtual lane center point sequence becomes a lane center point sequence of a virtual lane when a lane without lanes between a lane with an entry point and a lane with an exit point is connected by a virtual lane. . As for which approach point and exit point should be connected, lane L03 in Figure 5 has an additional lane and is a right-turn only lane, so the approach point located on lane L03 should be connected to lane L12 or L13 on the right-turn side of the intersection. Connect with the exit point located at .

At this time, by utilizing the information on the stop line F019 in the intersection shown in FIG. 13, the accuracy that the lane is connected to the right turn side increases, so it is good to have this information to prevent erroneous connections. Further, the entry point located on the lane L02 in FIG. 5 connects with the exit point located on the lane L04 or L05 on the left-turn side of the intersection and the straight lane L11.

FIG. 21 is a diagram illustrating an example of a method for generating a virtual lane center point sequence from intersection entry points and exit points. As shown in FIG. 21, a virtual lane center point sequence for connection uses an entry point F022 and an exit point F017 as control points for lane center point approximation. Then, using the intersection angle F034 of the point F031 where the extension line F033 of the lane to which the exit point F017 belongs and the extension line F032 of the lane to which the entry point F022 belongs intersect, it is approximated by a circular arc, clothoid curve, spline curve, Bezier curve, etc. do.

Note that each right/left turn method F018 and the stop line F019 in the intersection should be set as control points so as not to pass the right/left turn method F018 shown in FIG. 12 and to pass through the center of the stop line F019 in the intersection shown in FIG. It may be approximated as one of the following. Furthermore, as shown in FIG. 11, a virtual lane center point may be generated by utilizing the history of the position of the preceding vehicle F026. The preceding vehicle F026 may be equipped with a camera sensor S1 and a LiDAR sensor S2, and may obtain information from C2X and V2X.

As described above, with S05 to S08, even in laneless sections such as intersections and toll gates, by calculating the approach point, exit point, and intersection angle, the lane in the laneless section can be adjusted even in situations where there is little sensor information in the laneless section. This has the effect of making it possible to plan a level trip.

[Description of FIGS. 15 and 16] In FIG. 7, an example of the processing results of this embodiment for the sensor detection range F015 before the occurrence of the right turn lane at the intersection will be described.

First, an example of calculation by the sensor road/lane structure generation unit 105 in FIG. 3 will be explained.
Assume that the white line information in FIG. 10 and the roadside information in FIG. 9 are detected in the detection range F015 shown in FIG. 7 . In the example shown in FIG. 10, the white line information detects an increase in the number of right turn lanes, and also detects the white line on the left turn side of the intersection. Further, in the example shown in FIG. 9, the road edge information can detect the left turn side of the intersection and the right turn oncoming lane side.

Using the above data, first, through the processing from S01 to S04 in Figure 4, as shown in Figure 15, lanes 1 to 3 from the own vehicle position to the intersection, and lanes 4 and 5 on the left turn side of the intersection are The lane center point sequence can be detected.

Since the white line breaks before the intersection, the entry point to the intersection can be calculated for lane 2 and lane 3 in FIG. 15 by the process in S06 in FIG. Next, through the process of S07 in FIG. 4, the exit point from the intersection of lanes 4 and 5 can be calculated. Finally, through the process of S08 in FIG. 4, the lane center point sequence for lane 6 for connecting lane 2 and lane 4 in the no-lane section can be calculated. Furthermore, a lane center point sequence in lane 7 for connecting lane 2 and lane 5 can be calculated.

If each lane center point string can be generated, the left-right connection F030 of each lane in FIG. 15 can be generated, and the front-back connection F031 can be generated depending on the positional relationship of the lane center point strings. Then, areas R01, R02, R03, and R04 are generated as road information based on the increase/decrease of each lane and the boundary point of the intersection, and a connection F031 before and after them can be generated. Furthermore, although white line information has not been detected on the right-turn side of the intersection, since the corner on the right-turn side has been detected as road edge information, region R06 and its connection F032 can be generated.

Next, an example of calculation by the map synthesis unit 106 in FIG. 3 will be explained.
Assume that the information shown in FIG. 6 is provided as a simple map. According to this map, the intersection type is a crossroads, and in map link F002, there are two forward lanes and two oncoming lanes on the right/left turn side, and one forward lane and two oncoming lanes when going straight.

Comparing the calculation example of the sensor road/lane structure generation unit 105 with the simple map, the number of lanes and road connections in the information generated by the sensor road/lane structure generation unit 105 match the simple map, so the values shown in FIG. No correction is made in the error correction section. However, since the sensor road/lane structure generation unit 105 does not generate the road and lane information for going straight through the intersection and the lane information for the right-turn side of the intersection, the map synthesis unit 106 generates the road information F033 and the lane information for the right-turn side of the intersection, as shown in FIG. The lane 11 belonging thereto is synthesized with the result of the sensor road/lane structure generation unit 105.

Furthermore, on the right-turn side of the intersection, lane F034 (lane 12 and lane 13) is generated as a lane belonging to region R06 based on the information that there are two lanes of travel on the simple map. Then, an insufficient lane is generated for the area R04 within the intersection. Here, lane 8 connecting lane 2 and lane 11 is generated. Furthermore, lanes 9 and 10 connecting lanes 3, 12, and 13 are generated.

Note that by setting the reliability of the road information and lane information generated only from the simple map, it is possible to take safe measures to prevent the own vehicle speed from increasing too much as a control. For example, if lane 13 is the intended route, the information can be determined to have been generated from a simple map, so enter the intersection at low speed and gradually advance until you can begin to detect the white line information and road edge information on the right-turn side of the intersection. It becomes possible to take measures such as Providing only the information seen by the sensors means that until the information is seen, no control can be taken, which has the advantage of preventing a situation in which the vehicle cannot be moved. Furthermore, even if a right turn is not possible due to congestion or the like, there is an advantage that the vehicle can safely shift to an abnormal operation detected by the sensor, such as proceeding to a left turn.

In FIG. 7, an example of the processing results of this embodiment for the sensor detection range F020 when moving in the right turn lane at an intersection will be described.
When the vehicle moves to the right turn lane at the intersection, more white line information and road edge information can be detected. For example, with the white line information in FIG. 10, the white line information on the right turn side of the intersection can be detected in the detection range F020. Further, in the road edge information shown in FIG. 9, road edge information on the right and left turn sides of the intersection can be detected in the detection range F020.

Therefore, as shown in FIG. 16, the sensor road/lane structure generation unit 105 can generate lanes L12 and L13 that have not been detected so far, and can generate lanes L9 and L10 in the intersection connected to them. Furthermore, since no road edge is detected on the straight-ahead side of the intersection, region R05 can be generated. However, since no white line information has been detected, lane information cannot be generated.

On the other hand, when the map synthesis unit 106 processes the lane L11 in the region R05, since the simple map has one lane of travel, the lane L11 can be generated, and the lane L8 connected thereto can be generated.

[Description of FIG. 17] An example of the error correction section 308 in the map synthesis section 106 will be explained using FIG. 17.
Assume that in the detection range F020 of FIG. 7, two lanes, lane L11 and lane L14, are generated in the output of the sensor road/lane structure generation unit 105 shown in FIG. 17 for going straight through the intersection due to sensor erroneous detection. Judging from only the white line information, there are two oncoming lanes and one proceeding lane on the straight-ahead side of the intersection, so it is possible to misjudge the number of proceeding lanes.
At this time, using the information that the simple map has one lane of travel, lane L14 is treated as an error and excluded from the final output. This has the advantage that highly reliable information can be generated using more accurate information and reliability can be improved.

<Second embodiment>
FIG. 18 illustrates the configuration of an internal processing block of the driving support device in the second embodiment.
[composition]
FIG. 18 shows a configuration in which a high-precision map is additionally used as map information from the map/locator unit U1. 18 has a configuration in which a high-precision map information acquisition unit 131, a map road/lane structure generation unit 132, and an optimal road/lane structure selection unit 133 are added to the configuration of FIG. 2 in the first embodiment.

[Data Flow] Only the differences with respect to FIG. 2 will be explained, but the map/route information acquisition unit 101, position and orientation information acquisition unit 102, sensor information acquisition unit 103, sensor information integration unit 112, self-position and orientation estimation unit 104, and The sensor road/lane structure generation unit 105 and the flow of data to these blocks are the same as in FIG.

In FIG. 18, data 2031 from the map/locator unit U1 is input to the high-precision map information acquisition section 131, and data 2032 is output. Data 2032 from the high-precision map information acquisition unit 131 and data 2037 from the self-position and orientation estimation unit 104 are input to the map road/lane structure generation unit 132, and data 2033 is output.

Data 2005 from the map/route information acquisition unit 101, data 2018 from the self-position/orientation estimation unit 104, and data 2009 from the sensor road/lane structure generation unit 105 are input to the map synthesis unit, and data 2034 is output. . Note that the output data is the same as 2010 and 2012 in FIG.

Data 2034 from the map synthesis unit 106 and data 2033 from the map road/lane structure generation unit 132 are input to the optimal road/lane structure selection unit 133, and data 2035 and 2036 are output.

Data 2004 from the map/route information acquisition unit 101, data 2008 from the self-position/orientation estimation unit 104, and data 2035 from the optimal road/lane structure selection unit 133 are input to the route information addition unit 107, and data 2011 is output. .

Data 2036 from the optimal road/lane structure selection unit 133 and data 2011 from the route information addition unit 107 are input to the control unit 108, and data 2014 and 2015 are output.

[Explanation of functional blocks and data contents]
Unlike the data 2001, the data 2031 from the map/locator unit U1 is assumed to be high-precision map information, and is acquired by the high-precision map information acquisition unit 131, converted as necessary, and stored in memory. . In addition to the information of the simple map shown in FIG. 6, the high-precision map has lane level information, for example, information similar to F016, which is information representing the shape of each lane as shown in FIG. Furthermore, regarding the shape and location of map nodes, information with an accuracy of several centimeters is assumed for the simple map. Examples of conversion by the high-precision map information acquisition unit 131 include converting information in a global coordinate system such as latitude and longitude to a plane coordinate system. This facilitates distance calculation in the subsequent stage.

The data 2032 from the high-precision map information acquisition unit 131 assumes a high-precision map with lane-level information, and the data 2037 from the own position and orientation estimation unit 104 assumes the own vehicle position and orientation information. The road/lane structure generation unit 132 acquires data, transforms it as necessary, and stores it in memory. As an example of conversion by the map road /lane structure generation unit 132, a data format similar to the regions R01 to R06 and lanes L01 to L13 shown in FIG. Furthermore, data in the global coordinate system is converted into a vehicle relative coordinate system using the vehicle position and orientation information from the vehicle position and orientation estimation unit 104.

The data 2034 from the map synthesis unit 106 assumes a road and lane structure that has been synthesized with a simple map, and the data 2033 from the map road/lane structure generation unit 132 assumes a road and lane structure based on a high-precision map. , the optimal road/lane structure selection unit 133 considers the accuracy of data 2034 and data 2033, merges them into an optimal road/lane structure, and outputs data 2035 and 2036. If the reliability of the data 2034 based on sensor information is high, the data 2034 is merged with priority over the data 2033, but if the reliability is low, the data 2033 is merged with priority over the data 2044.

The data 2035 from the optimal road/lane structure selection unit 133 is in the same format as the data 2010 in FIG. 2, and the route information addition unit 107 generates a route from the road and lane information in the same manner as in FIG.

The data 2036 from the optimal road/lane structure selection unit 133 represents road and lane information whose accuracy has been improved using a high-precision map, and the control unit 108 implements the travel planning unit 109 and the trajectory planning unit 110. As a result, even when sensors are unreliable or simple maps are unreliable, using high-precision maps increases the reliability of information and enables safer control.

The data 2011 from the route information addition unit 107 that is input to the control unit 108 in FIG. 2 is assumed to have the same format as the data 2011 in FIG. This has the effect of reducing the number of man-hours required for switching between the systems shown in FIGS. 2 and 3.

Furthermore, it is assumed that the data 2012 from the map synthesis unit 106 that is input to the control unit 108 in FIG. 2 has the same format as the data 2036 from the optimal road/lane structure selection unit 133 in FIG.
The control unit 108 having the same logic can be operated in both the configuration of FIG. 2 and the configuration of FIG. 3, and there is an effect that the number of man-hours required for switching between the systems of FIG. 2 and FIG. 3 is reduced.

Although the embodiments of the present invention have been described in detail above, the present invention is not limited to the above-described embodiments, and various designs can be made without departing from the spirit of the present invention as described in the claims. Changes can be made. For example, the embodiments described above have been described in detail to explain the present invention in an easy-to-understand manner, and the present invention is not necessarily limited to having all the configurations described. Furthermore, it is possible to replace a part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. Furthermore, it is possible to add, delete, or replace some of the configurations of each embodiment with other configurations.

DESCRIPTION OF SYMBOLS 1... Own vehicle, 3... Driving support device, 101... Map/route information acquisition part, 102... Position and orientation information acquisition part, 103... Sensor information acquisition part, 105... Sensor Road/lane structure generation unit, 106...Map synthesis unit, 107...Route information addition unit, 109...Travel planning unit, 110...Trajectory planning unit, S...Sensor, U1... Map/locator unit, D... actuator

Claims (7)

a sensor information acquisition unit that acquires sensor information detected around the own vehicle by a sensor;
a sensor road/lane structure generation unit that generates road and lane structure information based on the sensor information;
a map synthesis unit that synthesizes map information selected based on the position information of the own vehicle and the road and lane structure information generated by the sensor road and lane structure generation unit;
The sensor road/lane structure generation unit includes:
generating a lane center point sequence of lanes that the road has based on road information included in the sensor information;
calculating entry points and exit points of a lane-free section of the road based on the lane center point sequence;
A lane center point sequence of a virtual lane for the vehicle to move from the approach point to the exit point is generated, the lane center point sequence of the lane and the lane center point sequence of the virtual lane are integrated, and the lane center point sequence of the virtual lane is integrated. Generate lane structure information indicating a connection relationship between the lanes with a lane section in between;
The map synthesis unit is
an error correction unit that compares the map information and the road and lane structure information and corrects information that does not match the map information in the road and lane structure information;
Generate lane structure information that is not included in the road and lane structure information based on the map information and the road and lane structure information corrected by the error correction unit, and generate lane structure information that is not included in the road and lane structure information. a lane information generation unit that synthesizes lane structure information with the lane structure information;
a road information generation unit that generates road information to which the lane structure information synthesized by the lane information generation unit belongs;
has
The road information generation unit generates the road information using a road and lane structure information having lane structure information that is a combination of lane structure information based on sensor information and lane structure information based on map information,
When the number of lanes on the road based on the sensor information and the number of lanes on the road based on the map information are different, the error correction unit is configured to change the number of lanes on the road based on the sensor information on the basis of the map information. Make corrections to match the number of lanes on the road.
A driving support device characterized by: The sensor road/lane structure generation unit includes:
determining lane increase/decrease boundaries of the road based on a lane center point sequence of the road;
generating lane structure information of the lane based on the position of a lane center point sequence of the road and lane increase/decrease boundaries of the road;
The road information of the road is calculated using the information on the lane increase/decrease boundary, data on which lane-free sections and other boundary information are assumed, and data on which the lane ID and the front/back/left/right connections between each lane are assumed. The driving support device according to claim 1, wherein the driving support device generates the following. It has a map/route information acquisition unit that acquires route information and information on a simple map used for route guidance from the map data ,
The map synthesis unit generates a simple lane center point sequence of the virtual lane based on information on the simple map, integrates the lane center point sequence of the lane and the simple lane center point sequence of the virtual lane, The driving support device according to claim 2, wherein the driving support device generates the road and lane structure information. A travel plan that determines the road on which the host vehicle should travel and the lane of the road based on the route information acquired by the map/route information acquisition unit and the road and lane structure information synthesized by the map synthesis unit. The driving support device according to claim 3, further comprising a portion. The driving support device according to claim 4 , wherein the driving planning unit outputs information about the road on which the own vehicle should travel and the lane of the road to a notification unit that notifies the driver of the own vehicle. . A trajectory plan including information on the trajectory and target speed of the vehicle is generated based on information on the road on which the vehicle should travel and lanes of the road, and the trajectory plan is transmitted to an actuator that controls the vehicle . The driving support device according to claim 4, further comprising a trajectory planning section that outputs the following . The driving support device according to claim 1, wherein the error correction unit determines whether the reliability of the map information is high or low, and performs the correction when the reliability is high.

Images