JP7844092B2 - Vehicle driving control device, autonomous vehicle, and vehicle driving control method - Google Patents

Description

This invention relates to a technology for controlling a vehicle in order to drive it toward a destination.

Patent Document 1 describes a technology that uses sensors mounted on the vehicle to detect markers (first and second physical features) placed on a cart that the vehicle is heading towards, thereby acquiring the positional relationship between the vehicle and the markers, and causing the vehicle to autonomously drive toward the cart. Such autonomous driving technology requires high-precision control of the vehicle's position relative to the markers. In contrast, Patent Document 2 describes a technology that estimates the vehicle's position by matching the marker positions with map data, and Patent Document 3 describes a technology that estimates the position and orientation by using markers having a moiré pattern.

US10168711 Japanese Patent Publication No. 2021-194995 Re-table No. 2018-135063

However, Patent Document 2 requires the preparation of map data, and Patent Document 3 requires the preparation of a light source environment for accurately reading moiré patterns. Therefore, there was a need for a technology that could easily and accurately control the position of a vehicle relative to a marker.

This invention has been made in view of the above-mentioned problems, and aims to enable simple and highly accurate control of the vehicle's position relative to a marker.

The vehicle driving control device according to the present invention comprises a position sensor attached to a vehicle, a marker position detection unit that uses the position sensor to detect the positions of N reference markers (where N is an integer of 2 or more) located at the vehicle's destination and acquires the detected position of each of the N reference markers, a storage unit that stores the reference position, which is the detected position of each of the N reference markers to be acquired by the marker position detection unit when the vehicle moves to the destination, for each of the N reference markers, and a drive control unit that controls a vehicle drive unit that drives the vehicle based on the difference between the reference position and the detected position of each of the N reference markers, thereby driving the vehicle toward the destination.

The autonomous vehicle according to the present invention comprises a vehicle body, a vehicle drive unit that drives the vehicle body, and the above-mentioned vehicle driving control device that controls the vehicle drive unit.

The vehicle driving control method according to the present invention comprises the steps of: detecting the positions of N reference markers (where N is an integer of 2 or more) located at the vehicle's destination using a position sensor attached to the vehicle, and acquiring the detected position of each of the N reference markers using a marker position detection unit; reading the reference position of each of the N reference markers from a storage unit that holds the reference position, which is the detected position of each reference marker to be acquired by the marker position detection unit, when the vehicle moves to the destination; and controlling a vehicle drive unit that drives the vehicle based on the difference between the reference position and the detected position of each of the N reference markers, thereby driving the vehicle toward the destination.

In the present invention (vehicle driving control device, autonomous driving vehicle, and vehicle driving control method) configured as described above, a position sensor is attached to the vehicle to detect the position of reference markers provided at the vehicle's destination relative to the vehicle. N reference markers (where N is an integer of 2 or more) are provided at the destination, and the marker position detection unit detects the positions of the N reference markers relative to the vehicle using the position sensor, acquiring the detected position of each of the N reference markers. Furthermore, the storage unit holds the reference position, which is the detected position of the reference marker to be acquired by the marker position detection unit when the vehicle moves to the destination, for each of the N reference markers. Then, the vehicle drive unit, which drives the vehicle, is controlled based on the difference between the reference position and the detected position of each of the N reference markers. In other words, the vehicle's movement is controlled based on the difference between the reference position of each of the N reference markers held in the storage unit and the detected position of each of the N reference markers detected using the position sensor. As a result, it is possible to control the vehicle's position relative to the markers simply and with high accuracy.

Furthermore, the vehicle travel control device may be configured such that the destination is an object to which N reference markers are attached, each of the N reference markers has a convex shape that protrudes from the object in a plan view, and the marker position detection unit acquires the position of the vertex of the convex shape of the reference marker as the detection position of the reference marker. In such a configuration, the detection position of the reference marker can be easily acquired by detecting the position of the vertex of the convex shape.

Furthermore, the vehicle driving control device may be configured such that the convex shape is V-shaped. In such a configuration, the detection position of the reference marker can be easily obtained by detecting the position of the vertex of the V-shape.

Furthermore, the vehicle driving control device may be configured such that the marker position detection unit includes: a point cloud data acquisition unit that scans the destination using a position sensor to acquire point cloud data showing the shape of an object and N reference markers provided at the destination; a minimum point extraction unit that extracts points from the point cloud data where the value of the Y coordinate is a minimum on the θ-Y plane, which is composed of a θ coordinate corresponding to the rotation direction in which the vehicle rotates and a Y coordinate corresponding to the straight-line direction in which the vehicle moves in a straight line; and a minimum point selection unit that acquires the positions of N minimum points that satisfy predetermined conditions from the M minimum points (M is a natural number greater than N) extracted by the minimum point extraction unit as points where each minimum value is a minimum, as the detection positions of each of the N reference markers. With such a configuration, the positions of N reference markers provided on the object can be accurately detected and the detection positions of each of the N reference markers can be acquired. Therefore, reference marks can be detected from a wide range of areas.

Furthermore, various predetermined conditions can be considered for obtaining the detection positions of each of the N reference markers from the positions of each of the M local minima. For example, the predetermined conditions may include the condition that the difference between the distance to the reference position and the distance to the local minima is less than a first threshold. The predetermined conditions may also include the condition that the difference between the distance between two of the N local minima and the distance between the reference positions of two of the N reference markers is less than a second threshold. Alternatively, the predetermined conditions may include the condition that the slope of the line indicated by multiple points in the point cloud data that are included in a predetermined target range ending at the local minima is within a predetermined slope range.

Furthermore, N may be 2. In this configuration, the vehicle's movement can be easily controlled based on the difference between the reference position of each of the two reference markers held in the memory unit and the detected position of each of the two reference markers detected using the position sensor.

According to the present invention, it becomes possible to control the position of a vehicle relative to a marker in a simple and highly accurate manner.

A schematic diagram showing an example of an autonomous vehicle according to the present invention and an object placed at the destination of the autonomous vehicle. A block diagram showing an example of the electrical configuration of an autonomous vehicle, as shown in Figure 1. A flowchart illustrating an example of autonomous vehicle driving control performed by the driving control unit. This figure schematically illustrates the actions performed according to the driving control shown in Figure 3. This figure schematically illustrates the actions performed according to the driving control shown in Figure 3. This figure schematically illustrates the actions performed according to the driving control shown in Figure 3. This figure schematically illustrates the actions performed according to the driving control shown in Figure 3. A flowchart showing an example of the process performed when detecting the marker position in Figure 3. A flowchart showing the calculations using the likelihood function in the flowchart of Figure 5. Figure 5 is a flowchart that schematically illustrates the operations performed on point cloud data.

Figure 1 is a schematic diagram showing an example of an autonomous vehicle according to the present invention and an object placed at the destination of the autonomous vehicle, and Figure 2 is a block diagram showing an example of the electrical configuration of the autonomous vehicle in Figure 1. Note that the dimensional relationships in Figure 1 are schematic and do not represent the actual ones. The same applies to the following figures.

The autonomous vehicle 1 is a so-called AGV (Automatic Guided Vehicle). As shown in Figure 1, this autonomous vehicle 1 comprises a vehicle body 11 and a plurality of wheels 12 that drive the vehicle body 11, and the autonomous vehicle 1 moves as the wheels 12 rotate. Note that the specific configuration for driving the vehicle body 11 is not limited to wheels 12, and for example, it may be a continuous track.

The autonomous vehicle 1 travels toward its destination 9 and stops upon reaching it. An object 91 is located at the destination 9, and upon reaching the destination 9, the autonomous vehicle 1 stops either docked with the object 91 or in close proximity to it.

This object 91 has a vertically erected wall 911, and two markers M1 and M2 are attached to the wall 911. Markers M1 and M2 are attached to the wall 911 with a horizontal gap between them and are at the same height. In a plan view (i.e., viewed from above in the vertical direction), markers M1 and M2 have a convex shape, particularly a V shape, protruding from the wall 911, and have vertices V1 and V2. In other words, markers M1 and M2 have left slopes Sl1 and Sl2 extending from the wall 911 toward vertices V1 and V2 to the left of vertices V1 and V2, and right slopes Sr1 and Sr2 extending from the wall 911 toward vertices V1 and V2 to the right of vertices V1 and V2. Note that right and left correspond to the right and left when the autonomous vehicle 1 approaches the object 91 and views the object 91 from the front.

As shown in Figure 2, the autonomous vehicle 1 has a drive motor 13 that drives the wheels 12 and a steering wheel 14 that changes the direction of the wheels 12. In other words, the drive motor 13 drives the wheels 12, causing them to rotate and the autonomous vehicle 1 to move. Also, the steering wheel 14 changes the direction of the wheels 12, thereby changing the direction in which the autonomous vehicle 1 moves.

The autonomous vehicle 1 is equipped with a LiDAR (Light Detection and Ranging) 2. The LiDAR 2 scans a predetermined range in front of the autonomous vehicle 1 to detect objects located in front of it and acquires point cloud data Dp representing the three-dimensional shape of those objects. In particular, the LiDAR 2 is used to detect the positions of two markers M1 and M2 provided at the destination 9.

Furthermore, the autonomous vehicle 1 includes a driving control unit 3 that controls the driving motor 13 and steering 14 based on point cloud data Dp acquired by LiDAR 2. This driving control unit 3 controls the driving motor 13 and steering 14 based on the positions of two markers M1 and M2 indicated by the point cloud data Dp, thereby driving the autonomous vehicle 1 toward the destination 9.

The driving control unit 3 includes a calculation unit 4 and a storage unit 5. The calculation unit 4 is a processor such as a CPU (Central Processing Unit), and the storage unit 5 is a storage device such as an SSD (Solid State Drive). The storage unit 5 holds reference position data Dr, which will be described later.

The calculation unit 4 has a marker position detection unit 41 that detects the positions of two markers M1 and M2 based on the point cloud data Dp acquired by LiDAR2. In particular, the marker position detection unit 41 detects the positions of the vertices V1 and V2 of markers M1 and M2 as the respective positions of markers M1 and M2. This marker position detection unit 41 includes a point cloud data acquisition unit 411, a minimum point extraction unit 412, and a minimum point selection unit 413. The point cloud data acquisition unit 411 acquires point cloud data Dp from LiDAR2. The functions of the minimum point extraction unit 412 and the minimum point selection unit 413 will be described later.

Furthermore, the calculation unit 4 includes a drive control unit 43. The drive control unit 43 controls the driving motor 13 and steering 14 based on the positions of the two markers M1 and M2 detected by the marker position detection unit 41, thereby driving the autonomous vehicle 1 toward the destination 9.

Figure 3 is a flowchart showing an example of autonomous vehicle driving control performed by the driving control unit, and Figures 4A to 4D are schematic diagrams showing the operations performed according to the driving control in Figure 3. In Figures 4A to 4D, an XY Cartesian coordinate system fixed to LiDAR2 (in other words, the vehicle body 11) is shown, where the Y coordinate indicates the position coordinate in the direction in which the autonomous vehicle 1 is moving straight (straight direction), and the X coordinate indicates the position coordinate in the direction perpendicular to the straight direction of the autonomous vehicle 1 (orthogonal direction).

LiDAR2 acquires point cloud data Dp representing the three-dimensional shape of surrounding objects by scanning in a rotational direction around a central axis parallel to the vertical. This point cloud data Dp is represented by polar coordinates, which are composed of a combination of distance and position in the rotational direction. Correspondingly, in Figures 4A to 4D, the detection positions p1 and p2 of markers M1 and M2 detected by LiDAR2 are shown in polar coordinates (r1, θ1) and (r2, θ2). Here, distances r1 and r2 are the distances from LiDAR2 to the detection positions p1 and p2 of markers M1 and M2, and angles θ1 and θ2 are the angles of the detection positions p1 and p2 of markers M1 and M2 relative to LiDAR2.

Furthermore, in Figures 4A to 4D, the reference positions P1 and P2 of markers M1 and M2 are shown in polar coordinates (R1, Θ1) and (R2, Θ2). Here, reference positions P1 and P2 are the detection positions p1 and p2 of markers M1 and M2 that the marker position detection unit 41 should acquire when the autonomous vehicle 1 moves to destination 9, and are included in the reference position data Dr. In other words, when the autonomous vehicle 1 reaches destination 9, the detection positions p1 and p2 of markers M1 and M2 coincide with the reference positions P1 and P2 of markers M1 and M2, respectively. In response to this, the marker position detection unit 41 and the drive control unit 43 of the marker position detection unit 41 read the reference position data Dr from the storage unit 5 as appropriate and execute the control described below.

Figures 4A to 4C show the state of LiDAR2 before it reaches destination 9, and Figure 4D shows the state of LiDAR2 when it reaches destination 9. Following the execution of the driving control shown in Figure 3, the autonomous vehicle 1 moves in the order shown in Figures 4A to 4D. For simplicity, the following explanation assumes that R1・sin(Θ1) = R2・sin(Θ2).

In step S101, the point cloud data acquisition unit 411 causes the LiDAR 2 to perform a scan and acquires point cloud data Dp from the LiDAR 2. Then, based on the point cloud data Dp, the point cloud data acquisition unit 411 acquires the detection positions p1 and p2 of the markers M1 and M2.

In step S102, the drive control unit 43 determines whether the amount of deviation of the detected positions p1 and p2 of markers M1 and M2 relative to the reference positions P1 and P2 of markers M1 and M2 in the rotation direction of polar coordinates is within an acceptable range. Specifically, the following conditions apply:
ΔΔy=||Y1-y1|-|Y2-y2||<ΔΔYth
A determination is made based on this. Here,
Y1: The Y-coordinate component of the reference position P1 of marker M1 (= R1・sin(Θ1))
y1: Y-coordinate component of the detection position p1 of marker M1 (= r1・sin(θ1))
Y2: The Y-coordinate component of the reference position P2 of marker M2 (= R2・sin(Θ2))
y2: Y-coordinate component of the detection position p2 of marker M2 (= r2・sin(θ2))
ΔΔYth: This is a predetermined threshold corresponding to the acceptable range.

In other words, if the distance ΔΔy shown in Figure 4A is less than the threshold ΔΔYth, it is determined that the deviation in the direction of rotation is within the acceptable range (YES in step S102), and the process proceeds to step S104. On the other hand, if the distance ΔΔy shown in Figure 4A is greater than or equal to the threshold ΔΔYth, it is determined that the deviation in the direction of rotation is outside the acceptable range (NO in step S102), and the process proceeds to step S103 and then to step S104. However, the method for evaluating the deviation in the direction of rotation is not limited to this example; the deviation in the direction of rotation may also be evaluated based on whether the angle between the straight line passing through the reference positions P1 and P2 of markers M1 and M2 and the straight line passing through the detection positions p1 and p2 of markers M1 and M2 is less than a predetermined threshold angle.

In this example, step S102 is executed in the state shown in Figure 4A, and the result is determined to be "NO". Therefore, in step S103, the drive control unit 43 controls the steering 14 according to the distance ΔΔy, rotating the autonomous vehicle 1 in the rotational direction so that the distance ΔΔy decreases to less than the threshold ΔΔYth. The result of executing step S103 from the state shown in Figure 4A is shown in Figure 4B. As shown in Figure 4B, the distance ΔΔy is substantially zero.

In step S104, the drive control unit 43 determines whether the amount of deviation of the detected positions p1 and p2 of markers M1 and M2 relative to the reference positions P1 and P2 of markers M1 and M2 in the Y direction (direction corresponding to the Y coordinate) is within the allowable range. Specifically, the following conditions apply:
Δy1=|Y1-y1|<Y1th
Δy2=|Y1-y1|<Y2th
A determination is made based on this. Here,
Y1th is a predetermined threshold corresponding to the acceptable range, and Y2th is a predetermined threshold corresponding to the acceptable range, so thresholds Y1th and Y2th are equal.

Here, distance Δy1 and distance Δy2 are equal. Therefore, if distance Δy1 shown in Figure 4B is less than the threshold Y1th, it is determined that the deviation in the Y direction is within the acceptable range (YES in step S104), and the process proceeds to step S106. On the other hand, if distance Δy1 shown in Figure 4B is greater than or equal to the threshold Y1th, it is determined that the deviation in the Y direction is outside the acceptable range (NO in step S104), and the process proceeds to step S105 and then to step S106.

In this example, step S104 is executed in the state shown in Figure 4B, and the result is determined to be "NO". Therefore, in step S105, the drive control unit 43 controls the travel motor 13 according to the distance Δy1, moving the autonomous vehicle 1 in the Y direction so that the distance Δy1 decreases to less than the threshold Y1th. The result of executing step S105 from the state shown in Figure 4B is shown in Figure 4C. As shown in Figure 4C, the distance Δy1 is virtually zero.

In step S106, the drive control unit 43 determines whether the amount of deviation of the detected positions p1 and p2 of markers M1 and M2 relative to the reference positions P1 and P2 of markers M1 and M2 in the X direction (direction corresponding to the X coordinate) is within the allowable range. In this example, the deviation to the X coordinate is effectively evaluated by evaluating the deviation in the rotational direction in polar coordinates instead of the X direction. Specifically, the following conditions apply:
Δθ1=|Θ1−θ1|<Θ1th
Δθ2=|Θ2−θ2|<Θ2th
A determination is made based on this. Here,
Θ1th: A predetermined threshold corresponding to the acceptable range. Θ2th: A predetermined threshold corresponding to the acceptable range, and thresholds Θ1th and Θ2th are equal.

In other words, if the angle Δθ1 shown in Figure 4C is less than the threshold Θ1th and the angle Δθ2 shown in Figure 4C is less than the threshold Θ2th, it is determined that the deviation in the X direction is within the acceptable range ("YES" in step S106), and the flowchart in Figure 3 is terminated. On the other hand, if the angle Δθ1 shown in Figure 4C is greater than or equal to the threshold Θ1th, or if the angle Δθ2 shown in Figure 4C is greater than or equal to the threshold Θ2th, it is determined that the deviation in the X direction is outside the acceptable range ("NO" in step S106), and step S107 is executed before the flowchart in Figure 3 is terminated.

In this example, step S106 is executed in the state shown in Figure 4C, and the result is determined to be "NO". Therefore, in step S107, the drive control unit 43 controls the travel motor 13 and steering 14 according to angles Δθ1 and Δθ2, moving the autonomous vehicle 1 to the X coordinate such that angles Δθ1 and Δθ2 decrease to less than thresholds Θ1th and Θ2th, respectively. The result of executing step S107 from the state shown in Figure 4C is shown in Figure 4D. As shown in Figure 4D, angles Δθ1 and Δθ2 are substantially zero.

Figure 5 is a flowchart showing an example of the process performed in the marker position detection in Figure 3, Figure 6 is a flowchart showing the calculations using the likelihood function in the flowchart of Figure 5, and Figure 7 is a schematic diagram showing the contents of the calculations performed on the point cloud data in the flowchart of Figure 5.

In step S201, the point cloud data acquisition unit 411 acquires point cloud data Dp from LiDAR2 in the manner described above. Figure 7 schematically shows point cloud data Dp, which is composed of multiple dots dt, each indicating a three-dimensional position.

In step S202, the minimum point extraction unit 412 extracts points (minimal points Ia, Ib, Ic) from the point cloud data Dp on the θ-Y plane, which is composed of a θ coordinate corresponding to the rotational direction in which the autonomous vehicle 1 rotates and a Y coordinate corresponding to the straight-line direction in which the autonomous vehicle 1 moves in a straight line, where the value of the Y coordinate is a minimum. In the example in Figure 7, three minimum points Ia, Ib, and Ic are extracted from the point cloud data Dp.

In step S203, the local minimum selection unit 413 determines whether the number M of local minimums Ia, Ib, and Ic extracted by the local minimum extraction unit 412 is greater than the number N of markers M1 and M2. In this example, the number of local minimums Ia, Ib, and Ic (3) is greater than the number N of markers M1 and M2 (2), so the result is "YES" and the process proceeds to step S204.

In step S204, the local minimum selection unit 413 generates combinations of selecting N local minimums (2 points) from M local minimums (3 points) Ia, Ib, and Ic. In step S205, the local minimum selection unit 413 calculates an evaluation value for one of the multiple combinations generated in step S204 using a likelihood function. Here, the likelihood function is a function that evaluates the degree of matching between the N local minimums included in the target combination and the markers M1 and M2, and outputs a smaller evaluation value the higher the degree of matching.

The likelihood function calculation algorithm shown in Figure 6 is executed by the local minimum selection unit 413. In step S301, the distance Δr1 is calculated for the leftmost local minimum and the distance Δr2 is calculated for the rightmost local minimum among the two local minimums included in a combination for which the likelihood function calculation is to be performed. Here, distance Δr1 is the absolute value of the difference between the distance from LiDAR2 to the leftmost local minimum and the distance R1 from LiDAR2 to the reference position P1, and distance Δr2 is the absolute value of the difference between the distance from LiDAR2 to the rightmost local minimum and the distance R2 from LiDAR2 to the reference position P2. For example, if the target is a combination consisting of local minimums Ia and Ib,
Δr1=|R1−ra|
Δr2=|R2−rb|
This is the case when the object is a combination consisting of local minima Ib and Ic,
Δr1=|R1−rb|
Δr2=|R2−rc|
This is the case when the object is a combination consisting of local minima Ia and Ic,
Δr1=|R1−ra|
Δr2=|R2−rc|
And the following conditions apply: Δr1 < R1th AND Δr2 < R2th
It is determined whether the condition in step S301 is met (step S301). Here, R1th and R2th are pre-set thresholds. If the condition in step S301 is not met ("NO"), the evaluation value C is set to infinity in step S306.

On the other hand, if the condition in step S301 is met ("YES"), the process proceeds to step S302. Then, the following condition is given between the distance d between the two local minimums included in one combination and the distance D between the two reference positions P1 and P2: Δd = |D - d| < ΔDth
It is determined whether the condition is met (step S302). Here, ΔDth is a predetermined threshold. For example, if the target is a combination consisting of local minima Ia and Ib,
Δd = |D - dab|
dab: This is the distance between two local minima Ia and Ib. When the object is a combination of local minima Ib and Ic,
Δd = |D - dbc|
dbc: This is the distance between two local minima Ib and Ic. When the object is a combination of local minima Ia and Ic,
Δd = |D - dac|
dac: This is the distance between the two local minima Ia and Ic. If the condition in step S302 is not met ("NO"), the evaluation value C is determined to be infinity in step S306.

On the other hand, if the condition in step S302 is met ("YES"), the process proceeds to step S303. Then, the slope m1 is calculated for the leftmost of the two local minima included in a given combination, and the slope m2 is calculated for the rightmost local minima. Here, the slope m1 is the slope of the straight line indicated by multiple dots dt included in a predetermined target range ending at the leftmost local minima (a range less than or equal to the width of the right slope Sr1 of the left marker M1), and the slope m2 is the slope of the straight line indicated by multiple dots dt included in a predetermined target range ending at the rightmost local minima (a range less than or equal to the width of the left slope Sl2 of the right marker M1). Note that the straight line indicated by multiple dots dt is, for example, a regression line for multiple dots dt. For example, if the target is a combination consisting of local minima Ia and Ib,
m1 = mar (the slope formed by the point cloud surrounding the right side of the local minimum Ia)
m² = mbl (the slope formed by the point cloud located to the left of the local minimum Ib)
This is the case when the object is a combination consisting of local minima Ib and Ic,
m1 = mbr (the slope formed by the point cloud surrounding the right side of the local minimum Ib)
m² = mcl (the slope formed by the point cloud located to the left of the local minimum Ic)
This is the case when the object is a combination consisting of local minima Ia and Ic,
m1 = mar (the slope formed by the point cloud surrounding the right side of the local minimum Ia)
m² = mcl (the slope formed by the point cloud located to the left of the local minimum Ic)
And the following conditions apply: M1thm < m1 < M1thp AND M2thm < m2 < M1thp
It is determined whether the condition is met (step S303). Here, M1thm, M1thp, M2thm, and M2thp are preset thresholds. If the condition in step S303 is not met ("NO"), the evaluation value C is set to infinity in step S306.

On the other hand, if the condition in step S303 is met ("YES"), the process proceeds to step S304. Then, of the two local minima included in a combination, the slope m1 of the left local minima and the slope m2 of the right local minima are given by the following condition: Δm = |m1 - m2| < ΔMth
It is determined whether the condition in step S304 is met (step S304). Here, ΔMth is a predetermined threshold. If the condition in step S304 is not met ("NO"), the evaluation value C is set to infinity in step S306.

On the other hand, if the condition in step S304 is met ("YES"), the process proceeds to step S305. Then, the following relationship is given: C = Δr1/R1 + Δr2/R2 + Δd/D
The evaluation value C is determined by this.

Returning to Figure 5, let's continue the explanation. Once the calculation of the evaluation value using the likelihood function in Figure 6 (step S205) is completed for all combinations ("YES" in step S206), the local minimum selection unit 413 selects the positions of the two local minimums of the combination with the smallest evaluation value as the detection positions p1 and p2 of the two markers M1 and M2, respectively (step S207). On the other hand, if "NO" is selected in step S203, the local minimum selection unit 413 selects the positions of the two local minimums extracted in step S202 as the detection positions p1 and p2 of the two markers M1 and M2, respectively (step S208). As a result, in the example in Figure 7, the positions of local minimums I1 and I3 are selected as the detection positions p1 and p2 of the two markers M1 and M2, respectively (steps S207, S208).

In the embodiment described above, a LiDAR 2 (position sensor) is attached to the autonomous vehicle 1 to detect the positions of markers M1 and M2 (reference markers) located at the destination 9 of the autonomous vehicle 1 relative to the autonomous vehicle 1. Two markers M1 and M2 are provided at the destination 9, and the marker position detection unit 41 detects the positions of the two markers M1 and M2 relative to the autonomous vehicle 1 using the LiDAR 2 and acquires the detected positions p1 and p2 of the two markers M1 and M2 respectively (step S102). Furthermore, the storage unit 5 stores the reference positions P1 and P2, which are the detected positions p1 and p2 of the markers M1 and M2 that the marker position detection unit 41 should acquire when the autonomous vehicle 1 moves to the destination 9, for each of the two markers M1 and M2. Then, the driving motor 13 and steering 14 (vehicle drive unit) that drive the autonomous vehicle 1 are controlled based on the difference between the reference positions P1 and P2 of the two markers M1 and M2 and the detected positions p1 and p2 of the two markers M1 and M2. In other words, the driving of the autonomous vehicle 1 is controlled based on the difference between the reference positions P1 and P2 of the two markers M1 and M2 held in the memory unit 5 and the p1 and p2 of the two markers M1 and M2 detected using LiDAR 2. As a result, the position of the autonomous vehicle 1 relative to the markers M1 and M2 can be controlled simply and with high precision.

Furthermore, the destination 9 is provided with an object 91 to which two markers M1 and M2 are attached, and each of the two markers M1 and M2 has a convex shape that protrudes from the object 91 in a plan view. In this case, the marker position detection unit 41 acquires the positions of the vertices V1 and V2 of the convex shape of the markers M1 and M2 as the detection positions p1 and p2 of the markers M1 and M2. With this configuration, the detection positions p1 and p2 of the markers M1 and M2 can be easily acquired by detecting the positions of the vertices V1 and V2 of the convex shape.

Furthermore, the convex shapes of markers M1 and M2 are V-shaped. In this configuration, the detection positions p1 and p2 of markers M1 and M2 can be easily obtained by detecting the positions of the vertices V1 and V2 of the V-shape.

Furthermore, the marker position detection unit 41 includes a point cloud data acquisition unit 411, a minimum point extraction unit 412, and a minimum point selection unit 413. The point cloud data acquisition unit 411 scans the destination 9 using LiDAR 2 to acquire point cloud data Dp that shows the shape of the object 91 and the two markers M1 and M2 provided at the destination 9. The minimum point extraction unit 412 extracts from the point cloud data Dp the minimum points I1, I2, and I3 whose Y coordinate values are minimums on the θ-Y plane, which is composed of a θ coordinate corresponding to the rotation direction in which the autonomous vehicle 1 rotates and a Y coordinate corresponding to the straight-line direction in which the autonomous vehicle 1 moves in a straight line. Furthermore, the minimum point selection unit 413 acquires the positions of two minimum points I1 and I3 that satisfy predetermined conditions (steps S301, S302, S303, S304) from among the three or more minimum points I1, I2, and I3 extracted by the minimum point extraction unit 412 as points with minimum values, as the detection positions p1 and p2 of the two markers M1 and M2. With this configuration, the positions of the two markers M1 and M2 provided on the object 91 can be accurately detected and the detection positions p1 and p2 of the two markers M1 and M2 can be acquired. Therefore, the markers M1 and M2 can be detected from a wide range of areas.

In the above embodiment, the autonomous driving vehicle 1 corresponds to an example of the "vehicle" and "autonomous driving vehicle" of the present invention, the vehicle body 11 corresponds to an example of the "vehicle body" of the present invention, the driving motor 13 and steering 14 constitute an example of the "vehicle drive unit" of the present invention, LiDAR 2 corresponds to an example of the "position sensor" of the present invention, LiDAR 2 and driving control unit 3 constitute an example of the "vehicle driving control device" of the present invention, the marker position detection unit 41 corresponds to an example of the "marker position detection unit" of the present invention, the point cloud data acquisition unit 411 corresponds to an example of the "point cloud data acquisition unit" of the present invention, the minimum point extraction unit 412 corresponds to an example of the "minimum point extraction unit" of the present invention, the minimum point selection unit 413 corresponds to an example of the "minimum point selection unit" of the present invention, and the drive control unit 43 is The following are examples of the present invention: the "drive control unit" corresponds to an example of the present invention, the storage unit 5 corresponds to an example of the "storage unit" of the present invention, the destination 9 corresponds to an example of the "destination" of the present invention, the object 91 corresponds to an example of the "object" of the present invention, the point cloud data Dp corresponds to an example of the "point cloud data" of the present invention, the two markers M1 and M2 correspond to an example of the "N reference markers (N is an integer of 2 or more)" of the present invention, the detection positions p1 and p2 correspond to an example of the "detection position" of the present invention, the reference positions P1 and P2 correspond to an example of the "reference position" of the present invention, the thresholds ΔR1th and ΔR2th correspond to an example of the "first threshold" of the present invention, the threshold ΔDth corresponds to an example of the "second threshold" of the present invention, and the ranges M1thm to M1thp and M2thm to M2thp correspond to an example of the "target range" of the present invention.

The present invention is not limited to the embodiments described above, and various modifications can be made to those described above without departing from the spirit of the invention. For example, the marker position detection unit 41 and the storage unit 5 may be built on a server computer separate from the autonomous vehicle 1, and the autonomous vehicle 1 may be controlled wirelessly from the server computer.

Furthermore, the number of markers M1 and M2, N, is not limited to two; it can be three or more.

Furthermore, the placement or shape of markers M1 and M2 may be changed as appropriate.

1. Autonomous vehicle (vehicle)
11... Vehicle body 13... Driving motor (vehicle drive unit)
14. Steering (vehicle drive unit)
2. LiDAR (position sensor, vehicle driving control system)
3. Driving control unit (vehicle driving control device)
41...Marker position detection unit 411...Point cloud data acquisition unit 412...Minimal point extraction unit 413...Minimal point selection unit 43...Drive control unit 5...Storage unit 9...Destination 91...Target object Dp...Point cloud data M1, M2...Markers (reference markers)
p1, p2... Detection position P1, P2... Reference position

Claims (9)

A position sensor attached to the vehicle,
A marker position detection unit detects the positions of N reference markers (where N is an integer of 2 or more) located at the vehicle's destination using the position sensor, and obtains the detected position of each of the N reference markers.
A storage unit that stores the reference position, which is the detection position of the reference marker to be acquired by the marker position detection unit, for each of the N reference markers when the vehicle moves to the destination,
The vehicle includes a drive control unit that controls a vehicle drive unit that drives the vehicle based on the difference between the reference position and the detection position of each of the N reference markers, thereby causing the vehicle to travel toward the destination .
At the destination, an object to which the N reference markers are attached is provided.
Each of the N reference markers has a convex shape that protrudes from the object in a plan view.
The marker position detection unit is a vehicle driving control device that acquires the position of the vertex of the convex shape of the reference marker as the detection position of the reference marker . The vehicle driving control device according to claim 1 , wherein the convex shape is V-shaped. The marker position detection unit is,
A point cloud data acquisition unit that scans the destination using the position sensor to acquire point cloud data showing the shape of the object and the N reference markers provided at the destination,
A minimum point extraction unit extracts points from the point cloud data where the value of the Y coordinate is a minimum on the θ-Y plane, which is composed of a θ coordinate corresponding to the direction of rotation in which the vehicle rotates and a Y coordinate corresponding to the direction of straight travel in which the vehicle moves in a straight line.
The vehicle driving control device according to claim 1, further comprising a minimum point selection unit that acquires the positions of N minimum points that satisfy predetermined conditions from among M minimum points (where M is a natural number greater than N) extracted by the minimum point extraction unit as points that each have a minimum value, as the detection positions of the N reference markers. The vehicle driving control device according to claim 3 , wherein the predetermined condition includes the condition that the difference between the distance to the reference position and the distance to the minimum point is less than a first threshold. The vehicle driving control device according to claim 3, wherein the predetermined condition includes the condition that the difference between the distance between two of the N local minimums and the distance between the respective reference positions of two of the N reference markers is less than a second threshold. The vehicle driving control device according to claim 3 , wherein the predetermined condition includes the condition that the slope of a straight line indicated by a plurality of points in the point cloud data that are included in a predetermined target range ending at the minimum point is within a predetermined slope range. The vehicle driving control device according to claim 1, wherein N is 2. The vehicle body and
A vehicle drive unit that drives the vehicle body,
An autonomous vehicle comprising a vehicle driving control device according to any one of claims 1 to 7 for controlling the vehicle drive unit. A step of detecting the position of N reference markers (where N is an integer of 2 or more) located at the vehicle's destination using a position sensor attached to the vehicle, and acquiring the detected position of each of the N reference markers using a marker position detection unit,
When the vehicle moves to the destination, the marker position detection unit reads the reference position, which is the detection position of the reference marker to be acquired by the marker position detection unit, from a storage unit that holds the reference position of each of the N reference markers,
The process includes controlling the vehicle drive unit that drives the vehicle using a drive control unit based on the difference between the reference position and the detection position of each of the N reference markers, thereby driving the vehicle toward the destination .
At the destination, an object to which the N reference markers are attached is provided.
Each of the N reference markers has a convex shape that protrudes from the object in a plan view.
The marker position detection unit is a vehicle driving control method that acquires the position of the vertex of the convex shape of the reference marker as the detection position of the reference marker .