JP7478696B2 - Road shape recognition device - Google Patents

Description

This disclosure relates to a road shape recognition device.

A known road shape recognition device recognizes road shapes by calculating a road model that represents the road shape with lines from the edge points of road division lines recognized by a camera. Patent Document 1 describes a technology that calculates a road model based on a previously recognized road model.

Patent No. 5559650

However, previously calculated road models do not necessarily correspond to the actual road shape. If a previously calculated road recognition model is incorrect, the road model cannot be calculated properly, and there is a risk that the road shape cannot be recognized. For this reason, there is a demand for technology that can recognize road shapes.

This disclosure has been made to solve the above-mentioned problems, and can be realized in the following forms:

According to a first aspect of the present disclosure, there is provided a road shape recognition device (110) mounted on a vehicle (10) having a camera (122). The road shape recognition device includes a surrounding information recognition unit (111) that recognizes surrounding information, the surrounding information recognition unit (111) being capable of recognizing image information that is the shape of each of one or more division lines of a road as at least a part of the surrounding information based on a surrounding image that is an image of the surroundings of the vehicle captured by the camera, a point sequence setting unit (112) that sets a point sequence that represents the shape of the road on which the vehicle is traveling based on the surrounding information recognized by the surrounding information recognition unit, and a first reliability of each point included in the point sequence set by the point sequence setting unit within a predetermined range. The road model calculation unit includes a first reliability setting unit (113) that sets a reliability by a numerical value , a road model calculation unit (114) that calculates a road model that represents the shape of the road on which the vehicle is traveling by a line, the road model calculation unit calculating a new road model based on the point sequence set by the point sequence setting unit, the first reliability set by the first reliability setting unit, and a road model previously calculated by the road model, a second reliability setting unit (115) that sets a second reliability of the road model by a numerical value within the predetermined range, and an output unit (117) that outputs the road model. The first reliability setting unit sets the first reliability of a point included in the point sequence in a section where the peripheral information includes the shape of only one of the left and right lane lines of the vehicle to be lower than the first reliability of a point included in the point sequence in a section where the peripheral information includes the shapes of both the left and right lane lines of the vehicle. The second reliability setting unit sets the second reliability such that the higher the first reliability of the point sequence in the section to which the road model corresponds, the higher the second reliability of the road model. The road model calculation unit determines a road model candidate based on the sequence of points, and in a section of the road ahead of the vehicle where a previously calculated road model and the road model candidate are separated by a threshold distance or more in the width direction of the road, if the first reliability at a point included in the sequence of points is lower than the second reliability in the previously calculated road model, calculates the road model based on the previously set sequence of points, and if the first reliability at a point included in the sequence of points is higher than the second reliability in the previously calculated road model, calculates the road model based on the sequence of points.

According to a second aspect of the present disclosure, there is provided a road shape recognition device mounted on a vehicle having a camera and a peripheral object sensor capable of measuring a distance to an object. The road shape recognition device includes a surrounding information recognition unit that can recognize surrounding information including image information which is the shape of one or more dividing lines of a road based on a surrounding image which is an image of the surroundings of the vehicle captured by the camera, and surrounding information including sensor information which is one or more of the shape of a roadside object and a movement trajectory of another vehicle based on an output of the surrounding object sensor; a point sequence setting unit that sets a sequence of points which represents the shape of a road on which the vehicle is traveling based on the surrounding information recognized by the surrounding information recognition unit; a first reliability setting unit that sets a first reliability of each point included in the sequence of points set by the point sequence setting unit to a numerical value within a predetermined range ; and a road model calculation unit that calculates a road model which represents the shape of the road on which the vehicle is traveling with lines, the road model calculation unit calculating a new road model based on the sequence of points set by the point sequence setting unit, the first reliability set by the first reliability setting unit, and a road model previously calculated by the road model; a second reliability setting unit that sets a second reliability of the road model to a numerical value within the predetermined range; and an output unit that outputs the road model. The first reliability setting unit sets the reliability of points included in a sequence of points in a section recognized based on the image information, which is the shape of a lane marking on only one of the left and right sides of the vehicle, to be lower than the reliability of points included in a sequence of points in a section recognized based on the sensor information. The second reliability setting unit sets the reliability in a section where the reliability of the sequence of points is low to be lower than the reliability in a section where the reliability of the sequence of points is high. The road model calculation unit determines a road model candidate based on the sequence of points, and calculates the road model based on the sequence of points set previously if the reliability of points included in the sequence of points is lower than the reliability of a previously calculated road model in a section of the road ahead of the vehicle where a previously calculated road model and the road model candidate are separated from each other by a threshold distance or more in a width direction of the road, and calculates the road model based on the sequence of points if the reliability of points included in the sequence of points is higher than the reliability of a previously calculated road model.

According to these types of road shape recognition devices, in sections where the previously calculated road model and the road model candidate are separated by a threshold distance or more in the road width direction, the road model calculation unit determines the sequence of points that serves as the basis for calculating the road model to be either the previously set sequence of points or the currently set sequence of points, depending on the reliability. In other words, in sections where the previously calculated road model and the road model candidate are separated, the road model calculation unit calculates the road model based on the sequence of points with high reliability. Therefore, the road shape can be recognized in accordance with the actual road shape.

FIG. 1 is a schematic diagram showing the configuration of an automatic driving control system. 13 is a flowchart showing a road shape recognition process. 11A and 11B are diagrams illustrating an example of peripheral information, a point sequence, and a road model candidate. FIG. 13 is a diagram showing an example of a road model calculated in the past. FIG. 11 is an explanatory diagram showing an example of calculation of a road model. FIG. 2 is an explanatory diagram illustrating an example of a road model. FIG. 11 is a schematic diagram showing a configuration of an automatic driving control system in a second embodiment. FIG. 11 is a diagram illustrating an example of sensor information.

A. First embodiment:
As shown in FIG. 1, the vehicle 10 includes an automatic driving control system 100. In this embodiment, the automatic driving control system 100 executes automatic driving of the vehicle 10. In this embodiment, the automatic driving control system 100 includes a road shape recognition device 110, a camera 122, a driving control unit 210, a driving force control ECU (Electronic Control Unit) 220, a braking force control ECU 230, and a steering control ECU 240. The road shape recognition device 110, the driving control unit 210, the driving force control ECU 220, the braking force control ECU 230, and the steering control ECU 240 are connected via an in-vehicle network 250. Note that the vehicle 10 may be driven not only by automatic driving but also by manual driving manually performed by a driver.

The camera 122 captures images of the surroundings of the vehicle 10.

The road shape recognition device 110 includes a surrounding information recognition unit 111, a first reliability setting unit 113, a point sequence setting unit 112, a road model calculation unit 114, a second reliability setting unit 115, a memory unit 116, and an output unit 117. The road shape recognition device 110 is composed of a microcomputer configured with a central processing unit (CPU), RAM, and ROM, and the functions of each of these units are realized by the microcomputer executing a program installed in advance. However, some or all of the functions of each of these units may be realized by a hardware circuit.

The surrounding information recognition unit 111 can recognize image information as at least a part of the surrounding information based on a surrounding image, which is an image captured by the camera 122. "Image information" refers to the shape of one or more road markings. "Surrounding information" refers to information related to the surroundings of the vehicle 10, including image information and sensor information described in the second embodiment. In this embodiment, the surrounding information recognition unit 111 recognizes the presence and positions of markings on the left and right sides of the road on which the vehicle is traveling, from the image captured by the camera 122. The surrounding information recognition unit 111 recognizes the shape of the markings, etc., from this information, for example, using a Kalman filter or the least squares method.

The point sequence setting unit 112 sets a point sequence representing the shape of the road on which the vehicle 10 is traveling, based on the surrounding information recognized by the surrounding information recognition unit 111. In this embodiment, the point sequence setting unit estimates the coordinate position of the point sequence based on the shape of the road division lines recognized by the surrounding information recognition unit 111.

The first reliability setting unit 113 sets a first reliability for each point included in the sequence of points set by the sequence of points setting unit. The "first reliability" indicates the degree of certainty of the shape of the road indicated by each point. The higher the first reliability, the higher the possibility that the shape can be trusted. The first reliability setting unit 113 can set the first reliability according to the surrounding information used to set the sequence of points. In this embodiment, the first reliability setting unit 113 sets the first reliability of a point included in the sequence of points in a section where the surrounding information includes the shape of only one of the lane lines on either the left or right side of the vehicle to a lower value than the first reliability of a point included in the sequence of points in a section where the surrounding information includes the shape of both the lane lines on the left and right sides of the vehicle.

The road model calculation unit 114 calculates a road model that uses lines to represent the shape of the road on which the vehicle is traveling. The road model calculation unit 114 calculates the road model based on the point sequence set by the point sequence setting unit, the first reliability set by the first reliability setting unit, and a road model previously calculated by the road model. Details of the calculation of the road model will be described later.

The second reliability setting unit 115 sets the second reliability of the road model as a numerical value within the same range of magnitude as the range of the first reliability set by the first reliability setting unit. The "second reliability" indicates the degree of certainty of the shape of the road represented by the road model. The higher the second reliability, the higher the possibility that the shape can be trusted. The second reliability setting unit 115 sets the second reliability so that the higher the first reliability of the point sequence of the section corresponding to the road model, the higher the second reliability of the road model.

The memory unit 116 stores the set point sequence, the calculated road model, and the second reliability of the road model.

The output unit 117 outputs the road model calculated by the road model calculation unit 114 to the driving control unit 210 etc. via the in-vehicle network 250.

The driving control unit 210 is composed of a microcomputer configured with a central processing unit (CPU), RAM, and ROM, and realizes an automatic driving function by having the microcomputer execute a pre-installed program. The driving control unit 210 controls the driving force control ECU 220, the braking force control ECU 230, and the steering control ECU 240, for example, using a sequence of points representing the road shape set by the point sequence setting unit 112.

The driving force control ECU 220 is an electronic control device that controls actuators that generate the driving force of the vehicle, such as the engine. When the driver drives manually, the driving force control ECU 220 controls the power source, which is the engine or electric motor, according to the amount of accelerator pedal operation. On the other hand, when autonomous driving is performed, the driving force control ECU 220 controls the power source according to the required driving force calculated by the driving control unit 210.

The braking force control ECU 230 is an electronic control device that controls the brake actuator that generates the braking force of the vehicle. When the driver drives manually, the braking force control ECU 230 controls the brake actuator according to the amount of operation of the brake pedal. On the other hand, when the vehicle is driven automatically, the braking force control ECU 230 controls the brake actuator according to the required braking force calculated by the driving control unit 210.

The steering control ECU 240 is an electronic control device that controls the motor that generates the steering torque of the vehicle. When the driver drives manually, the steering control ECU 240 controls the motor in response to the operation of the steering wheel to generate an assist torque for the steering operation. This allows the driver to operate the steering wheel with a small amount of force, and realizes steering of the vehicle. On the other hand, when autonomous driving is performed, the steering control ECU 240 performs steering by controlling the motor in response to the required steering angle calculated by the driving control unit 210.

The road shape recognition process shown in FIG. 2 is a series of processes in which the road model calculation unit 114 recognizes the road shape of the route along which the vehicle 10 is traveling. This process is repeatedly executed by the road shape recognition device 110 while the vehicle 10 is traveling, and is repeatedly executed every 100 ms, for example.

In step S100, the surrounding information recognition unit 111 recognizes the surrounding information. In this embodiment, the surrounding information recognition unit 111 recognizes image information as at least a part of the surrounding information based on a surrounding image, which is an image captured by the camera 122.

In step S110, the point sequence setting unit 112 sets a point sequence that represents the road shape using points based on the surrounding information recognized in step S100.

In step S120, the first reliability setting unit 113 sets the first reliability of each point included in the sequence of points set in step S110 based on the surrounding information recognized in step S100.

In step S130, the road model calculation unit 114 determines a road model candidate based on the sequence of points set in step S110. A "road model candidate" is a candidate for a line that represents the road shape.

In step S140, the road model calculation unit 114 calculates a road model. In this embodiment, in a section of the road ahead of the vehicle where the previously calculated road model and the road model candidate determined in step S130 are separated by a threshold distance or more in the road width direction, the road model calculation unit 114 calculates a new road model according to the relationship between the first reliability of the points included in the sequence of points set in step S120 in that section and the second reliability of the previously calculated road model in that section. If the first reliability of the points included in the sequence of points is lower than the second reliability of the previously calculated road model, the road model calculation unit 114 calculates a new road model based on the sequence of points set in step S110. On the other hand, if the first reliability of the points included in the sequence of points is higher than the second reliability of the previously calculated road model, the road model calculation unit 114 calculates a new road model based on the sequence of points set in step S110.

In step S150, the second reliability setting unit 115 sets the second reliability to the road model calculated in step S140. This process may be performed before step S140. In this case, the road model calculation unit 114 sets the second reliability to the new road model calculated in step S140 in the previous road shape recognition process.

As shown in FIG. 3, the surrounding information recognized in step S100 (see FIG. 2) of the road shape recognition process is the shape of the right-hand demarcation line LnR and the left-hand demarcation line LnL of the lane Ln1 in which the vehicle 10 is traveling. The shape of the demarcation line LnR is indicated by a sequence of points Pr1 to Pr7. The shape of the demarcation line LnL is indicated by a sequence of points Pl1 to Pl4. The surrounding information recognition unit 111 recognizes the image information by converting the captured image into a bird's-eye view. Therefore, the farther away from the vehicle 10 is the image information recognized by the surrounding information recognition unit 111, the more likely it is that the shape of the actual demarcation lines will deviate due to the gradient of the road and the pitching of the vehicle 10.

In FIG. 3, points P1 to P7 included in the sequence of points set by the sequence of points setting unit 112 based on the peripheral information in step S110 (see FIG. 2) are represented by crosses. For example, for a section where the peripheral information includes the shape of the left and right demarcation lines of lane Ln1, the sequence of points setting unit 112 sets a sequence of points at the center in the width direction of lane Ln1 of points at positions corresponding to the sequence of points Pr1 to Pr4 indicating the shape of demarcation line LnR and points Pl1 to Pl4 indicating the shape of demarcation line LnL. Also, for a section where the peripheral information includes the shape of only one of the left and right demarcation lines of lane Ln1, the sequence of points is set based on points Pr5 to Pr7 indicating the shape of demarcation line LnR. Therefore, as shown in FIG. 3, the sequence of points set by the sequence of points setting unit 112 is curved toward the demarcation line LnR.

In FIG. 3, the road model candidate MC1 determined by the road model calculation unit 114 in step S130 (see FIG. 2) is represented by a dashed line. The road model calculation unit determines the road model candidate based on the point sequence. Therefore, the road model candidate MC1 is curved toward the lane marking LnR.

The surrounding information recognized in step S100 (see FIG. 2) in the previous road shape recognition process shown in FIG. 4 is the shape of the right-hand demarcation line LnR and the left-hand demarcation line LnL of the lane Ln1 in which the vehicle 10 is traveling. As shown in FIG. 4, the shape of the demarcation line LnR is indicated by a sequence of points Pr1 to Pr7. The shape of the demarcation line LnL is indicated by a sequence of points Pl1 to Pl7.

In Figure 4, points P01 to P07 included in the point sequence set by the point sequence setting unit 112 based on the surrounding information in step S110 (see Figure 2) of the previous road shape recognition process are represented by triangle marks. In Figure 4, the road model M0 calculated by the road model calculation unit 114 in step S140 (see Figure 2) of the previous road shape recognition process is represented by a dashed line.

As shown in FIG. 5, the calculation of the road model in section S2, among the sections of the road ahead of the vehicle, where the distance d between the previously calculated road model M0 and the road model candidate MC1 is equal to or greater than the threshold distance in the road width direction, will be described. The points included in the point sequence in section S2 are points included in the point sequence in the section whose surrounding information includes the shape of only one of the lane markings on either the left or right side of the vehicle. The previously calculated road model in section S2 was calculated based on points P05 to P07 (see FIG. 4) included in the point sequence in the section whose surrounding information includes the shape of both the lane markings on the left and right sides of the vehicle. Therefore, the first reliability of points P05 to P07 included in the point sequence is lower than the second reliability of the previously calculated road model M0. Therefore, the road model calculation unit 114 calculates the road model in section S2 based on the point sequence previously set.

For a section S1 of the road ahead of the vehicle where the previously calculated road model M0 and the road model candidate MC1 are not separated by a threshold distance or more in the road width direction, the road model calculation unit 114 calculates a road model based on the sequence of points set in step S110 of the current road recognition process. That is, in step S140 (see FIG. 2), the road model calculation unit adopts the road model candidate MC1 determined in step S130 as the road model.

As shown by the solid line in Figure 6, the road model calculation unit 114 calculates the road model M1.

According to the road shape recognition device 110 of the present embodiment described above, in a section S2 where the previously calculated road model M0 and the road model candidate MC1 are separated by a threshold distance or more in the width direction of the road, the road model calculation unit 114 determines the point sequence that serves as the basis for the calculation of the road model M1 to be either the point sequence previously set or the point sequence currently set, depending on the reliability. For example, if the previously calculated road model M0 was calculated based on the shapes of both the left and right demarcation lines, and the currently determined road model candidate MC1 was determined based on the shape of only the left or right demarcation lines, the road model calculation unit 114 calculates the road model M1 in the section S2 based on the point sequence previously set. Also, if the previously calculated road model M0 was calculated based on the shapes of only the left and right demarcation lines, and the currently determined road model candidate MC1 was determined based on the shapes of both the left and right demarcation lines, the road model calculation unit 114 calculates the road model M1 in the section S2 based on the point sequence currently set. That is, in the section S2 where the previously calculated road model M0 and the road model candidate MC1 diverge, the road model calculation unit 114 calculates the road model M1 based on a highly reliable point sequence. Therefore, the road shape can be recognized in accordance with the actual road shape.

B. Second embodiment:
The configuration of the autonomous driving control system 100B of the second embodiment shown in FIG. 7 is different from that of the first embodiment in that a surrounding object sensor 124 is provided, but the other configurations are the same.

The surrounding object sensor 124 measures the distance to objects in the vicinity of the vehicle 10. Examples of the surrounding object sensor 124 include object sensors that use reflected waves, such as laser radar, millimeter wave radar, Lidar (Light Detection and Ranging), and ultrasonic sensors.

In this embodiment, the surrounding information recognition unit 111 recognizes sensor information, which is one or more of the shape of roadside objects and the movement trajectories of other vehicles, as part of the surrounding information based on the output of the surrounding object sensor 124.

The surrounding information shown in FIG. 8 is sensor information that indicates the movement trajectory of another vehicle 20 traveling in front of the vehicle 10. In FIG. 8, the road model M2 is represented by a dashed line.

In this embodiment, the first reliability setting unit 113 sets the first reliability of the points included in the sequence of points of the section recognized based on image information that is the shape of the lane markings on only one of the left and right sides of the vehicle to be lower than the first reliability of the sequence of points of the section recognized based on the sensor information.

For example, the calculation of the road model will be explained using as an example a case where the previously calculated road model was road model M2 (see FIG. 8) and the currently determined road model candidate is road model candidate MC1 (see FIG. 5). If the previously calculated road model M2 (see FIG. 8) was calculated based on the shapes of roadside objects detected by Lidar and the movement trajectories of other vehicles, and the currently determined road model candidate MC1 (see FIG. 5) was determined based on the shapes of the lane markings on only one side, the left or right, the road model calculation unit 114 calculates the road model M2 in that section based on the previously set point sequence.

The calculation of the road model will be explained using as an example a case where the previously calculated road model was road model candidate MC1 (see FIG. 5) and the currently determined road model candidate is road model M2 (see FIG. 8). If the previously calculated road model, road model candidate MC1 (see FIG. 5), was calculated based on the shape of the lane markings on only one side, and the currently determined road model candidate, road model M2 (see FIG. 8), was determined based on the shape of roadside objects detected by Lidar and the movement trajectories of other vehicles, the road model calculation unit 114 calculates the road model M2 in that section based on the currently set point sequence.

According to the first reliability setting unit 113 of this embodiment described above, the first reliability setting unit 113 sets the first reliability of the points included in the sequence of points of the section recognized based on image information that is the shape of the lane markings on only one of the left and right sides of the vehicle 10 to be lower than the first reliability of the sequence of points of the section recognized based on the sensor information. A road model calculated based on the shape of roadside objects detected by Lidar and the movement trajectory of other vehicles 20 conforms more closely to the actual road shape than a road model calculated based on a sequence of points set based on the shape of the lane markings on only one of the left and right sides. Therefore, according to this embodiment, the road shape can be recognized in accordance with the actual road shape.

C. Other embodiments:
(C1) In the above-described embodiment, the road model calculation unit 114 calculates the road model M1 based on the point sequence set in the current road recognition process in the section S1 where the previously calculated road model M0 and the road model candidate MC1 are not separated by a threshold distance or more in the road width direction. Not limited to this, the road model calculation unit 114 may calculate the road model M1 based on the point sequence newly set by the point sequence setting unit 112 in the current road shape recognition process and the point sequence previously set. The road model calculation unit 114 can calculate the road model M1 based on the newly set point sequence and the point sequence previously set by the point sequence setting unit 112 by weighting the newly set point sequence and the point sequence previously set by the point sequence setting unit 112 according to the time when the point sequence setting unit 112 set the point sequence. For example, the road model calculation unit 114 can weight the time when the point sequence is set closer to the current time. This allows the road model to be calculated based on the newly set point sequence rather than the point sequence previously set.

(C2) In the second embodiment, the first reliability setting unit 113 may set the first reliability of a point included in the sequence of points in section S2 recognized based on image information that is the shape of the lane markings on both the left and right sides of the vehicle 10 to be higher than the first reliability of the sequence of points in section S2 recognized based on sensor information.

The present disclosure is not limited to the above-described embodiments, and can be realized in various configurations without departing from the spirit of the present disclosure. For example, the technical features in the embodiments corresponding to the technical features in each form described in the Summary of the Invention column can be replaced or combined as appropriate to solve the above-described problems or to achieve some or all of the above-described effects. Furthermore, if a technical feature is not described as essential in this specification, it can be deleted as appropriate.

10...vehicle, 20...other vehicles, 100, 100B...automatic driving control system, 110...road shape recognition device, 111...surrounding information recognition unit, 112...point sequence setting unit, 113...first reliability setting unit, 114...road model calculation unit, 115...second reliability setting unit, 116...storage unit, 117...output unit, 122...camera, 124...surrounding object sensor, 210...driving control unit, 220...driving force control ECU, 230...braking force control ECU, 240...steering control ECU, 250...in-vehicle network

Claims (4)

A road shape recognition device (110) mounted on a vehicle (10) having a camera (122),
a surrounding information recognition unit (111) that recognizes surrounding information, the surrounding information recognition unit (111) being capable of recognizing image information that is the shape of each of one or more division lines of a road as at least a part of the surrounding information based on a surrounding image that is an image of the surroundings of the vehicle captured by the camera;
a point sequence setting unit (112) that sets a point sequence representing a shape of a road on which the vehicle is traveling based on the surrounding information recognized by the surrounding information recognition unit;
a first reliability setting unit (113) that sets a first reliability of each point included in the sequence of points set by the sequence of points setting unit to a numerical value within a predetermined range ;
a road model calculation unit (114) for calculating a road model representing, by lines, the shape of a road on which the vehicle is traveling, the road model calculation unit calculating a new road model based on the point sequence set by the point sequence setting unit, the first reliability set by the first reliability setting unit, and a road model previously calculated by the road model;
a second reliability setting unit (115) that sets a second reliability of the road model to a numerical value within the predetermined range;
an output unit (117) that outputs the road model,
the first reliability setting unit sets a first reliability of a point included in a sequence of points in a section where the peripheral information includes a shape of only one of the left and right lane lines of the vehicle to be lower than a first reliability of a point included in a sequence of points in a section where the peripheral information includes the shapes of both the left and right lane lines of the vehicle;
the second reliability setting unit sets the second reliability such that the higher the first reliability of the point sequence in the section corresponding to the road model, the higher the second reliability of the road model;
The road model calculation unit
determining a road model candidate based on the sequence of points;
In a section of the road ahead of the vehicle, a previously calculated road model and the road model candidate are separated from each other by a threshold distance or more in a width direction of the road.
if the first reliability of a point included in the sequence of points is lower than the second reliability of a previously calculated road model, calculate the road model based on a previously set sequence of points;
a road shape recognition device that calculates the road model based on the point sequence when the first reliability of a point included in the point sequence is higher than the second reliability of a previously calculated road model. 2. The road shape recognition device according to claim 1,
The vehicle has a surrounding object sensor (124) capable of measuring a distance to an object around the vehicle;
the surrounding information recognition unit is capable of recognizing, based on an output of the surrounding object sensor, sensor information that is one or more of a shape of a roadside object and a movement trajectory of another vehicle as part of the surrounding information;
A road shape recognition device, wherein the first reliability setting unit sets the reliability of a point included in a sequence of points of a section recognized based on the image information, which is the shape of a dividing line on only one of the left or right sides of the vehicle, to be lower than the reliability of a sequence of points of a section recognized based on the sensor information. 3. The road shape recognition device according to claim 1,
the road model calculation unit calculates a road model based on the sequence of points newly set by the point sequence setting unit and the sequence of points previously set in a section where the previously calculated road model and the road model candidate are not separated from each other by the threshold distance or more in a width direction of the road,
A road shape recognition device in which the calculation of the road model based on the newly set sequence of points and the previously set sequence of points is performed by weighting the newly set sequence of points and the previously set sequence of points according to the time when the sequence of points was set by the sequence of points setting unit. A road shape recognition device mounted on a vehicle having a surrounding object sensor capable of measuring a distance between a camera and an object,
a surrounding information recognition unit that can recognize surrounding information including image information of the shape of one or more dividing lines of a road based on a surrounding image that is an image of the surroundings of the vehicle captured by the camera, and surrounding information including sensor information of one or more of the shape of a roadside object and a movement trajectory of another vehicle based on an output of the surrounding object sensor;
a point sequence setting unit that sets a point sequence representing a shape of a road on which the vehicle is traveling based on the surrounding information recognized by the surrounding information recognition unit;
a first reliability setting unit that sets a first reliability of each point included in the sequence of points set by the sequence of points setting unit to a numerical value within a predetermined range ;
a road model calculation unit that calculates a road model that represents, by lines, a shape of a road on which the vehicle is traveling, and calculates a new road model based on the point sequence set by the point sequence setting unit, the first reliability set by the first reliability setting unit, and a road model previously calculated by the road model;
a second reliability setting unit that sets a second reliability of the road model to a numerical value within the predetermined range;
an output unit that outputs the road model,
the first reliability setting unit sets a reliability of a point included in a sequence of points of a section recognized based on the image information, which is a shape of a lane marking on only one of the left and right sides of the vehicle, to be lower than a reliability of a point included in a sequence of points of a section recognized based on the sensor information;
the second reliability setting unit sets a reliability in a section in which the reliability of the point sequence is low to be lower than a reliability in a section in which the reliability of the point sequence is high;
The road model calculation unit
determining a road model candidate based on the sequence of points;
In a section of the road ahead of the vehicle, a previously calculated road model and the road model candidate are separated from each other by a threshold distance or more in a width direction of the road.
If the reliability of a point included in the sequence of points is lower than the reliability of a previously calculated road model, the road model is calculated based on a previously set sequence of points;
a road shape recognition device that calculates the road model based on the point sequence when the reliability of a point included in the point sequence is higher than the reliability of a previously calculated road model.

Images