JP7382134B2 - Digital map and how to use it to drive autonomously - Google Patents

Description

The present invention relates to a digital map that can be treated as numerical values by a computer, and an automatic driving method using this digital map.

The Japan Digital Road Map Association provides digital maps (digital maps) that can be treated as numerical values on a computer. This digital map is widely used in navigation systems.

The Japanese government is aiming to introduce "semi-automated driving systems" to "fully automated driving systems" in the early to late 2020s. To achieve this goal, it is essential to create digital maps (dynamic maps) that contain three-dimensional spatial information. In addition, it is required to have not only information that does not change over time, such as roads and buildings, but also information that changes from moment to moment, such as information on traffic jams and the progress of nearby vehicles.
The Society of Automotive Engineers of Japan reports on the evaluation and traffic flow analysis of traffic accident simulators. The Cabinet Office (SIP) is moving not only to construct high-precision maps for self-location estimation and route identification, but also to construct dynamic maps as intelligent road traffic information databases for all vehicles.

Patent Document 1 describes that when a vehicle is driven automatically, data indicating that another automatically driven vehicle has switched to manual operation is received as a digital map indicating a switching position.

Patent Document 2 states that the conditions for allowing a car to operate automatically include that a structural separation part exists on at least one side of the current roadway for the car, and that the car lane has a minimum lane width. The radius of curvature of the lanes is greater than a predetermined threshold, there are no protrusions and depressions that would substantially limit the reach of the surrounding sensor, the number of lanes does not change, and there are no tunnels. , there are no buildings such as toll plazas on the roadway, there are no expressway interchanges, there are no traffic jams, there is no traffic information about dangerous situations, and there is no information about the existence of construction sites. It mentions things.

Patent Document 3 describes a method for correcting the position of a vehicle using the global navigation satellite system GNSS, which measures the position of the own vehicle. , thereby calculating a second position of the vehicle, identifying surrounding objects whose positions can be matched in a digital map, and using sensors to detect the actual distance between the vehicle and the object. However, the content of calculating the calculated distance between the second position and the object is disclosed.

Further, Non-Patent Document 1 discloses a manual that shows standard verification and evaluation procedures for evaluating the functional level and traffic situation reproduction ability of various traffic accident prediction simulation systems.

Japanese Patent Application Publication No. 2017-117456 Special table 2017-530045 publication Special table 2017-513020 publication

“Traffic Accident Simulation System Verification Manual” Society of Automotive Engineers of Japan, Traffic Accident Prediction Simulation Certification Review Committee https://www.jase.or.jp/tops/topics/1040/1040-1A

Conventional automatic driving control that does not use digital maps considers all surrounding obstacles detected by recognition sensors as subject to speed limits. For this reason, unnecessary speed control occurs, such as slowing down when detecting vehicles in lanes that do not overlap with the own vehicle's driving lane or pedestrians on the sidewalk.

In addition, in conventional automatic driving control, even if a lane change is to be performed, there is no information for determining an adjacent lane, so it is not possible to determine whether surrounding detected objects will interfere with the lane change.

Furthermore, with conventional automatic driving control, it is not possible to determine whether a vehicle can continue in its own lane or whether it must cross over into an adjacent lane to avoid a vehicle parked on the shoulder of the road.

Furthermore, when turning (turning right) at an intersection, it is difficult to judge whether or not it is possible to pass by passing an oncoming vehicle.

Patent Documents 1 to 3 listed above all describe content related to digital maps, but they do not include how to determine the own lane during automatic driving, what determines that it is an adjacent lane, and even more. There is no disclosure or suggestion as to whether or not the object is an obstacle that obstructs the movement of the vehicle.
Furthermore, Non-Patent Document 1 only discloses verification of traffic accident simulation, but does not disclose anything about how to create a digital map.

In order to solve the above problems, in the digital map for automatic driving according to the present invention, white line data and road boundary data are defined along a determined driving route, and the white line data includes, in addition to the actual white lines, Portions where white lines do not exist are supplemented with virtual white line data, and the road boundary data has a configuration in which, in addition to real boundaries, portions where no boundary lines exist are supplemented with virtual boundary line data. did.
Intersections are examples of areas where white line data and boundary line data do not exist on the map.

Further, in the automatic driving method according to the present invention using the digital map described above, when driving along a determined driving route, the first white line existing on the left and right with respect to the normal direction of the driving route is The area between the two lanes is determined to be the own lane, and the area between the first white line and the second white line is determined to be the adjacent lane based on the normal direction of the driving route. The area between existing road boundary lines is set as a recognition target range in which obstacles are detected by sensors.

According to the digital map and the automatic driving method using this digital map according to the present invention, when driving automatically, it is possible to accurately determine the own lane in which the own vehicle should travel, including intersections.

Therefore, even if there is an obstacle in front of the vehicle, the speed will not be reduced even if the obstacle is far from the vehicle's own lane, allowing efficient speed control.

In addition, it is possible to accurately determine the lane adjacent to the own lane in which the own vehicle is traveling, and to recognize obstacles on the adjacent lane, allowing for efficient timing, for example, changing lanes after waiting for another vehicle traveling in the adjacent lane to pass. Can be done.

Furthermore, by using sensors to detect obstacles between the left and right road boundaries, the system recognizes unnecessary objects, such as pedestrians walking on safe sidewalks, and avoids problems such as slowing down. It can be resolved.

Illustration of the procedure for creating a digital map Diagram showing a digital map with white line data and road boundary line data obtained from map data Diagram showing a digital map with supplemented white lines and road boundaries Diagram showing an example of numerical representation of a digital map (a) is the own lane, (b) is the adjacent lane, and (c) is a diagram explaining the recognition target range in which obstacles are detected by the sensor. Explanatory diagram of automatic driving method using digital map

To create a digital map for use in automated driving, white line data and road boundary data are acquired based on available digital maps, as shown in Figure 1.

For white line data, position data of white lines existing on the route is acquired (step 1) and defined as white lines for determining lanes. In addition, positional data of guardrails, median strips, etc. on the left and right sides of the route is acquired (step 2), and the range in which speed limits are recognized is defined as the road boundary.
The map shown in FIG. 2 shows white lines and road boundaries based on these data.

Since there are no white lines along the route at intersections, etc., virtual white lines are used to help autonomous driving recognize its own lane (Step 3). Similarly, even if there are no guardrails or median strips, road boundaries in areas that do not need to be recognized as subject to speed restrictions are supplemented (Step 4). The map shown in FIG. 3 is obtained by supplementing white lines and road boundaries in this way.

The white lines and road boundaries are expressed using longitude and latitude. FIG. 4 is an example of road boundaries expressed in terms of longitude and latitude.

Figure 5(a) explains the method of recognizing the own lane, where the own lane is defined between the first white lines w1 and w2 on the left and right, with the normal n of the route p that the own vehicle is traveling as a reference. It is recognized as the own lane.

Fig. 5(b) explains the method for recognizing adjacent lanes. I recognize that.

Figure 5(c) explains the recognition target range in which obstacles are detected by the sensor equipped on the vehicle, and the recognition target range is between the boundary lines b1 and b2 on the left and right, with the normal n of the route p as a reference. shall be.

Next, an example of automatic driving using the digital map will be explained according to the flowchart of FIG. 4.
First, the driving position of the own vehicle is specified based on information such as GPS, and the own lane is recognized using the above-described method from the driving position and white line information on the digital map (Step 5).

Next, adjacent lanes are recognized using the method described above from the driving position of the own vehicle and the position information of the white lines on the digital map (Step 6). Further, while the vehicle is running, the recognition target range (between the boundaries b1 and b2) is recognized (step 7), and it is detected whether or not there is an obstacle in this recognition target range.

Here, the position of the obstacle detected by the recognition sensor is in relative coordinates with respect to the own vehicle, so based on information such as GPS, the position of the obstacle detected by the recognition sensor is set to the same coordinate system as within the own lane, within the adjacent lane, and within the road boundary. Then, it is identified where the obstacle is located within the own lane, within the adjacent lane, or within the road boundary (Step 8).

Obstacles within the own lane, adjacent lanes, or road boundaries are targeted for control.
Specifically, when driving in a straight line, if there is an obstacle within the own lane (including when it is partially covered), the vehicle will adjust the speed (decelerate or stop), and when there is no obstacle within the own lane. continues to run.

When changing lanes to the adjacent lane, if there is an obstacle in the adjacent lane (including partially covered), adjust the timing of the lane change, and if there is no obstacle in the adjacent lane, continue changing lanes. I do.

If an obstacle is detected when you are not driving in a straight line or changing lanes, such as when turning at an intersection or determining whether or not to proceed, if the obstacle is located within the own lane, an adjacent lane, or within the road boundary, the control target shall be.

After avoiding the obstacle, if the running has not ended, the process returns to step 5, and if the running has ended, the program ends.

w1, w2, w3...White line b1, b2...Boundary line p...Route n...Normal line

Claims (1)

An automatic driving method using a computer based on numerical information of a digital map in which road boundaries and white lines on the road are treated as numerical values, the digital map having a driving route as a reference range for recognition to detect obstacles. Between the road boundary line data that exists on the left and right, the first two white line data that exists on the left and right with the driving route as a reference, and the one white line data that is closer to the center of the road among these two white line data. white line data for determining adjacent lanes, and furthermore, for locations including intersections where white line data and road boundary data along the route do not exist, virtual white line data and roads that connect interrupted white lines and road boundaries are defined. Boundary data is supplemented and defined, and the driving position of the own vehicle is specified based on the GPS information, and the own lane, adjacent lanes, and road boundaries are recognized from the identified driving position and the information on the digital map, and the driving position of the own vehicle is determined based on the GPS information. Sensors installed in the vehicle detect the presence or absence of obstacles between road boundaries, and according to instructions from the computer, the vehicle decelerates or stops if there is an obstacle within the own lane, and if there is no obstacle within the own lane. The automatic driving method is characterized by driving the vehicle as it is, adjusting the timing when changing lanes to an adjacent lane and when there is an obstacle in the adjacent lane, and changing lanes as it is when there is no obstacle.

Images