Autonomous lawnmower
The autonomous lawn mower uses imagers to gather grass data for reinforcement learning, optimizing mowing operations and ensuring efficient, high-quality cutting and post-processing.
Patent Information
- Application Number
- DE112022006887
- Authority / Receiving Office
- DE · DE
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2022-03-23
- Publication Date
- 2025-01-02
AI Technical Summary
Conventional autonomous lawn mowers lack the ability to autonomously adjust their operation based on real-time lawn conditions, such as grass state and weather, leading to suboptimal mowing performance.
The autonomous lawn mower employs forward and reverse imagers to capture information about uncut and cut grass, including moisture content, number, thickness, and color, which a control unit uses for reinforcement learning to determine optimal mowing operations, adjusting parameters like blade speed and height, and issuing post-processing instructions.
Enables the autonomous lawn mower to perform optimal mowing operations tailored to current conditions, ensuring clean cuts and minimizing uncut grass, with the ability to identify and address post-processing needs.
Smart Images

Figure 00000000_0000_ABST
Abstract
Description
Technical field
[0001] The present invention relates to an autonomous lawn mower. background
[0002] Lawn mowers equipped with a measuring device for measuring the activity level of the lawn are conventionally known (see, for example, Patent Document 1). This lawn mower collects information about the lawn's condition based on the results of a measurement with the measuring device and generates a map. Therefore, this type of lawn mower is considered suitable for contributing to lawn management. Citation list
[0003] Patent Document 1: Unexamined Japanese Patent Application, Publication No. 2019-88252 Disclosure of the invention Problems to be solved by the invention
[0004] In recent years, there has been a demand for autonomous lawn mowers to improve work efficiency. Lawn conditions are not always consistent due to various factors such as weather; therefore, autonomous lawn mowers should be adapted to optimally mow the lawn under current conditions.
[0005] However, the above conventional technology assumes that the optimal operation based on the created map is performed by a human using the map, and does not disclose how the lawnmower autonomously performs optimal lawnmowing operation.
[0006] Therefore, the present invention aims to provide an autonomous lawn mower capable of autonomously performing an optimal lawn mowing operation tailored to the conditions of the lawn. Means to solve the problems
[0007] The autonomous lawn mower (e.g., autonomous lawn mower 1, which will be described later) according to the present invention includes: a lawn mower body (e.g., lawn mower body 2, which will be described later) that performs a lawn mowing operation during autonomous driving; a control unit (e.g., control unit 5, which will be described later) that controls the lawn mower body based on a received operation content; a forward imager (e.g., forward imager 3, which will be described later) oriented forward in a traveling direction of the lawn mower body; and a backward imager (e.g., backward imager 4, which will be described later) oriented backward in a traveling direction of the lawn mower body.When a reverse imaging result detected by the reverse imager is within a predetermined range of conditions, the control unit determines the operation content based on a forward imaging result detected by the forward imager (3).
[0008] (2) In the autonomous lawn mower described in (1), the forward imaging result may include at least information of a state of uncut grass (e.g., grass Ga described later).
[0009] (3) In the autonomous lawn mower described in (2), the information of a state of uncut grass may include at least information of a moisture content of grass.
[0010] (4) In the autonomous lawn mower described in (3), the information of a state of uncut grass may further include information of at least one of a number of grasses and a thickness of grass, and the control unit may prioritize the information of a moisture content of grass over the information of the number of grasses and / or the thickness of grass when determining the operation content.
[0011] (5) In the autonomous lawn mower described in (4), the information of a state of uncut grass may further include the information of both the number of grass and the thickness of grass, and the control unit may prioritize the information of a thickness of grass over the information of the number of grass when determining the operation content.
[0012] (6) In the autonomous lawn mower described in any one of (2) to (5), the reverse imaging result may include at least information of the amount of grass (tufts of grass clippings Gc1 described later) ejected from the lawn mower body.
[0013] (7) In the autonomous lawn mower described in (6), the reverse imaging result may further include information of a state of mowed grass (e.g., grass Gb, which will be described later), and the control unit may prioritize the information of the amount of grass over the information of a state of mowed grass when determining the operation content.
[0014] (8) In the autonomous lawn mower described in (7), the information of a state of mowed grass may include information of a color of grass and a length unevenness of grass, and the control unit may prioritize information from a change in the color of unmowed grass (Ga) included in the forward imaging result to the color of mowed grass included in the reverse imaging result over the information of a length unevenness of grass when determining the operation content.
[0015] According to (1), the autonomous lawn mower can determine an operation content for optimal lawn mowing based on the forward imaging result acquired by the forward imager and the reverse imaging result acquired by the reverse imager; therefore, the autonomous lawn mower can autonomously perform optimal lawn mowing. Therefore, an autonomous lawn mower capable of autonomously performing optimal lawn mowing can be achieved.
[0016] According to (2), the autonomous lawn mower can perform lawn mowing operation based on the operation content taking into account the state of the uncut grass, which includes the forward imaging result.
[0017] According to (3), the autonomous lawn mower can perform a lawn mowing operation based on the operation content taking into account the moisture content of the uncut grass included in the forward imaging result.
[0018] According to (4), the autonomous lawn mower can perform a lawn mowing operation based on the more appropriate operation content tailored to the moisture content of the uncut grass included in the forward imaging result.
[0019] According to (5), the autonomous lawn mower can perform lawn mowing operation based on the more suitable operation content tailored to the thickness of the uncut grass.
[0020] According to (6), the autonomous lawn mower can perform a lawn mowing operation based on the operation content taking into account the amount of ejected grass included in the reverse imaging result.
[0021] According to (7), the autonomous lawn mower can perform a lawn mowing operation based on the more appropriate operation content tailored to the amount of ejected grass included in the reverse imaging result.
[0022] According to (8), the autonomous lawn mower can perform a lawn mowing operation based on the more appropriate operation content tailored to the change of a color of mowed grass included in the reverse imaging result. Brief description of the drawings Fig. 1 is a schematic side view illustrating an autonomous lawn mower; Fig. Figure 2A is a diagram illustrating proper grass mowing; Fig. Figure 2B is a diagram illustrating the condition of properly mown grass; Fig. Figure 3A is a diagram illustrating improper grass mowing; Fig. Figure 3B is a diagram illustrating white-tipped grass as a result of improper mowing; Fig. 4 is a plan view showing grass left behind during mowing; Fig. 5 is a functional block diagram of a control unit; Fig. 6 is a functional block diagram of an operation content determination unit; Fig. Figure 7 is a diagram showing an example of a reward table; Fig. 8 is a diagram showing an example of a setting value adjustment table; Fig. 9 is a diagram showing an example of an operation route from the start to an end of the operation by the autonomous lawn mower; Fig. 10 is a flowchart illustrating an example of a control operation for the autonomous lawn mower when determining an operation content; Fig. 11 is a flowchart illustrating an example of a control operation for the autonomous lawn mower when determining post-processing; Fig. 12A is a flowchart showing an example of the control operation of post-processing by the autonomous lawn mower; Fig. 12B is a flowchart showing an example of the control operation of post-processing by the autonomous lawn mower; Fig. 13 shows a post-processing by the autonomous lawnmower in the work area; Fig. 14 shows an example of the direction of travel during post-processing by the autonomous lawnmower; and Fig. 15A is a diagram illustrating the concept of overlap; and Fig. Figure 15B is a diagram illustrating the directions of travel during re-processing by the autonomous lawn mower when there is an error in the overlap width. Preferred mode for carrying out the invention Outline of the configuration of an autonomous lawn mower
[0023] As in Fig. 1, an autonomous lawn mower 1 comprises a lawn mower body 2, a forward imaging device 3, a reverse imaging device 4, a control unit 5, and a storage unit 6. The autonomous lawn mower 1 comprises a pair of front wheels 11 and a pair of rear wheels 12 and travels autonomously when mowing the lawn on vegetation within a working area, driven by a traction motor 21 integrated into the lawn mower body 2. In addition to the above, the autonomous lawn mower 1 comprises a steering mechanism for changing the direction of travel, rechargeable traction batteries, etc., which are not included in Fig. 1 are shown.
[0024] In describing the autonomous lawn mower 1 of the present embodiment, the direction indicated by the arrow in Fig. 1 indicates the direction of travel while mowing the lawn. Therefore, the direction shown in Fig. 1 of the autonomous lawn mower 1 of the present embodiment in the direction of travel forward to the left and backward to the right. The width direction of the autonomous lawn mower 1 of the present embodiment is a direction perpendicular to the surface of the paper in Fig. 1.
[0025] The lawn mower body 2 accommodates a cutting unit 24 at the bottom, which includes a lawn mower blade part. The specific structure of the blade part is not particularly limited. In the present embodiment, the blade part consists of a pair of cutting blades 241 arranged side by side in the width direction of the lawn mower body 2. The cutting blades 241 rotate at a predetermined speed around a rotation axis approximately perpendicular to the ground, driven by a blade motor 22 integrated into the lawn mower body 2, to mow the grass Ga in front of them in the traveling direction. When the autonomous lawn mower 1 mows the lawn properly, the grass Gb is trimmed to a predetermined length after passing the autonomous lawn mower 1.
[0026] The cutting unit 24 is vertically movable relative to the lawnmower body 2 and is driven by a unit lifting motor 23 integrated into the lawnmower body 2. The vertical movement of the cutting unit 24 adjusts the height of the cutting blades 241 relative to the ground, thereby ensuring the correct length of the mowed grass Gb.
[0027] The cutting unit 24 has a discharge opening 242 opening at the rear of the lawnmower body 2. The discharge opening 242 discharges and distributes the grass clippings Gc, which are mown by the cutting blades 241, backward in the direction of travel by the air flow generated by the rotation of the cutting blades 241. The sufficiently distributed grass clippings Gc can be reused as soil fertilizer.
[0028] The forward imaging device 3 is oriented forward in the direction of travel of the lawnmower body 2. In particular, the forward imaging device 3 is mounted slightly downward on the upper front part of the lawnmower body 2 to enable imaging of the uncut grass Ga in the direction of travel during mowing.
[0029] As a specific example of the forward imaging device 3, a camera including an image sensor capable of color image recognition of the grass Ga may be used. The forward imaging device 3 may be a moisture-sensing camera capable of measuring the moisture content of the grass Ga. For example, a moisture-sensing camera using near-infrared spectrometry may be used as the moisture-sensing camera.
[0030] The forward imaging device 3 may include two or more types of cameras, such as a camera with image sensors and a near-infrared spectrometer moisture meter. The forward imaging device 3 captures images of the unmowed grass Ga in front of it in the direction of travel and sends the forward imaging results, including the condition of the grass Ga, to the control unit 5.
[0031] The reverse imaging device 4 is oriented rearward in the direction of travel of the lawnmower body 2. Specifically, the reverse imaging device 4 is mounted slightly downward on the upper rear part of the lawnmower body 2 to enable imaging of the grass Gb in the opposite direction of travel during mowing.
[0032] As a specific example, a camera including an image sensor capable of color image recognition of the grass Gb can be used for the reverse imaging device 4. The reverse imaging device 4 captures images of the mowed grass Gb in the opposite direction of travel and sends the reverse imaging result, including the condition of the grass Gb, to the control unit 5.
[0033] The following describes the state of the grass Ga included in the forward imaging result and the state of the grass Gb included in the reverse imaging result.
[0034] The condition of the grass Ga contained in the forward imaging result may include the number (density), thickness, color, moisture content, etc. of the grass Ga. These conditions of the grass Ga can be detected by image processing of the forward imaging result.
[0035] The number and thickness of the grass Ga are relevant to the ease of mowing the grass Ga. For example, the grass Ga may be difficult to cut if it is numerous or thick. In such cases, adjustments such as reducing the traveling speed of the autonomous lawn mower 1 or increasing the rotation speed of the cutting blades 241 may be necessary. Therefore, information about the number and thickness of the grass Ga, as well as the condition of the grass Ga, can be used as a basis for determining the speed of the autonomous lawn mower 1 and the rotation speed of the cutting blades 241 for more appropriate lawn mowing.
[0036] The color of the grass Ga is relevant for the assessment of the final condition of the mown grass Gb. For example, as in Fig. 2A, the grass Ga is neatly mowed by the cutting blades 241 when the speed of the cutting blades 241 is appropriate. Consequently, the mowed grass Gb was, as shown in Fig. 2B, cleanly cut to a predetermined length, resulting in a minimal color difference between the uncut grass Ga and the cut grass Gb. As shown in Fig. However, as shown in Figure 3A, if the speed of the cutting blades 241 is slow or the sharpness of the cutting blades 241 decreases due to wear, the cutting blades 241 may press the grass Ga downward, making it difficult to mow the grass Ga cleanly. As a result, as shown in Fig. 3B, fibers are exposed as white tips Gb1, which remain at the mowed edge of the mowed grass Gb. Consequently, the color of the grass Gb generally becomes whitish and differs significantly from the color of the grass Ga. Therefore, the color of the grass Ga can be used, for example, as a criterion for determining the appropriate rotation speed of the cutting blades 241.
[0037] The moisture content of the grass Ga refers to the amount of moisture Ga adhering to the grass. The moisture content of the grass Ga can be measured using a moisture-sensing camera. The moisture content of the grass Ga can also be measured by detecting the light reflected from the grass Ga using a camera with an image sensor. The moisture content of the grass Ga can also include external input information such as rainfall or irrigation. The moisture adhering to the grass contributes to an increase in the mass of the grass. For example, at a higher moisture content, the grass is more prone to bending. Consequently, the grass Ga is unlikely to stand upright and is therefore more likely to remain unmown.Therefore, the moisture content of the grass Ga can be used, for example, as a criterion for determining the appropriate rotation speed of the cutting blades 241 to generate updrafts that can help straighten the grass Ga that has been bent over by moisture.
[0038] The grass condition Gb contained in the reverse imaging result is the final processing condition of the mowed grass Gb. The final processing condition of the grass Gb may include, among other things, the appearance of unmowed grass, the color of the grass Gb, the amount of fallen grass, and the length of the grass Gb (mowing height, length unevenness). These final processing conditions of the grass Gb can be detected by image processing of the reverse imaging result.
[0039] Fallen grass refers to the grass clippings Gc that accumulate as tufts on the grass Gb without being dispersed when ejected backwards in the direction of travel of the autonomous lawnmower 1. As a result, as shown in Fig. 4, the grass clippings Gc accumulated as tufts Gc1 are scattered on the grass Gb toward the rear in the traveling direction of the autonomous lawn mower 1. Since sunlight and moisture under the tufts of grass clippings Gc1 are less likely to reach the grass Gb, this grass Gb may die. Therefore, a large amount of grass forming the tufts of grass clippings Gc1 must be removed immediately.
[0040] The control unit 5 comprehensively controls the operation of the autonomous lawn mower 1. As in Fig. 5, the control unit 5 includes a motor control unit 51, an operation content setting unit 52, an operation content determining unit 53, a post-processing determining unit 54, a positioning unit 55, and a communication unit 56.
[0041] The motor control unit 51 generates operating commands for various motors based on the set values of the operating content set in the operating content setting unit 52. The motor control unit 51 sends the generated operating commands to various motors, such as the traction motor 21, the blade motor 22, and the work lift motor 23, thereby controlling the drive of the various motors. This realizes the mowing of the lawn by the autonomous lawn mower 1.
[0042] The operation content setting unit 52 sets the operation content required for mowing the lawn of the autonomous lawn mower 1. The operation content includes parameters for operating the various parts of the autonomous lawn mower 1. The operation content includes information about the traveling direction and speed of the autonomous lawn mower 1, the rotation speed of the cutting blades 241, and the height of the cutting unit 24. The operation content set in the operation content setting unit 52 is rewritable and is adjusted based on the information sent from the operation content determination unit 53.
[0043] As in Fig. As shown in Figure 6, the operation content determination unit 53 determines and outputs new operation contents for achieving better results in lawn mowing based on both the forward imaging result acquired by the forward imaging device 3 and the reverse imaging result acquired by the reverse imaging device 4. The operation content determination unit 53 identifies the state of the grass Gb included in the reverse imaging result acquired by the reverse imaging device 4, and, when the state of the grass Gb is within a predetermined range, determines a new operation content based on the state of the grass Ga included in the forward imaging result acquired by the forward imaging device 3.
[0044] The operation content determination unit 53 can determine new operation contents, for example, through reinforcement learning. Reinforcement learning involves learning the lawn mowing (action) performed by the autonomous lawn mower 1 to maximize future rewards. Specifically, the operation content determination unit 53 performs lawn mowing (action) based on determined operation contents and selects an action that improves the condition of the unmowed grass Ga detected in the forward imaging result to achieve a better condition of the mowed grass Gb detected in the reverse imaging result.
[0045] In reinforcement learning by the operation content determination unit 53, as a result of mowing the lawn based on the operation content set in the operation content setting unit 52, the reward is assigned a positive value when the condition of the mowed grass Gb is better than the condition of the unmowed grass Ga, and the reward is assigned a negative value when the condition of the mowed grass Gb is worse than the condition of the unmowed grass Ga. Specifically, for example, a reward table as shown in Fig. 7, is preset in the operation content determination unit 53. Based on the reward table, the operation content determination unit 53 evaluates, for example, the condition of the uncut grass Ga in the forward imaging result as "normal," "slightly poor," or "poor" on a three-level scale, and evaluates the condition of the mowed grass Gb in the reverse imaging result as "good," "medium," or "poor" on a three-level scale. The operation content determination unit 53 calculates either a positive or negative reward value based on the combinations of the above values.
[0046] As in Fig. As shown in Figure 7, the reward values are weighted. For example, the case where the grass condition Ga in the forward imaging result is "poor" and the grass condition Gb in the reverse imaging result is "good" indicates that the operation content of mowing the lawn is extremely good. Therefore, in this combination, the reward value is given a higher weight than in other combinations. Conversely, the case where the grass condition Ga in the forward imaging result is "normal" and the grass condition Gb in the reverse imaging result is "poor" indicates that the operation content of mowing the lawn is poor. Therefore, the reward value in this combination is given a lower weight than in other combinations.
[0047] This method of calculating rewards is an example and not limited to this. Another example of a method of calculating rewards may include quantifying the forward images and backward images and calculating (e.g., adding) these numerical values, thereby calculating the reward values. For example, the condition of the grass Ga included in the forward imaging result may be quantified as "normal: 0 points," "slightly poor: 3 points," or "poor: 5 points," and the condition of the mowed grass Gb included in the backward imaging result may be quantified as "good: 5 points," "medium: 3 points," or "poor: -1 point." The reward is calculated by adding the numerical values of the forward and backward images.For example, even if the grass condition Ga included in the forward imaging result is "poor: 5 points," but the mowed grass condition Gb included in the reverse imaging result is "good: 5 points," the total score is 10 points, which is the highest reward. Conversely, even if the grass condition Ga included in the forward imaging result is "normal: 0 points" and the mowed grass condition Gb included in the reverse imaging result is "poor: -1 point," the total score is -1 point, which is the lowest reward.
[0048] Based on the condition of the unmowed grass Ga included in the forward imaging result, the operation content of mowing the grass Ga, the condition of the mowed grass Gb included in the reverse imaging result, and the reward values calculated therefrom, the operation content determination unit 53 determines a new operation content as a new action for the current state, for example, through reinforcement learning. The operation content determination unit 53 establishes a connection between the condition of the grass Ga, the condition of the grass Gb, the operation content of mowing such grass, and the rewards, and accumulates them as experience to determine a new operation content.
[0049] The operation content, which is newly determined by the operation content determination unit 53 through reinforcement learning, includes adjustments such as increasing or decreasing the rotation speed of the cutting blades 241, raising or lowering the height of the cutting unit 24, or accelerating or decelerating the traveling speed of the autonomous lawn mower 1, which aims to obtain a higher positive reward compared to the current operation content. In particular, the operation content determination unit 53 is preset with a setting value adjustment table, such as shown in Fig. 8. For example, if the current operation content of the blade motor 22 is set to, for example, "3" (b3), the operation content determination unit 53 adjusts the new operation content to, for example, "2" (b2) or "4" (b4). The new operation content is sent to the operation content setting unit 52, where the lawn mowing operation content recorded in the operation content setting unit 52 is updated to an operation content that enables more appropriate lawn mowing.
[0050] The operation content determination unit 53 determines the operation content, and the operation content setting unit 52 makes adjustments during a predetermined control cycle after the autonomous lawn mower 1 starts mowing. Based on the lawn mowing operation content rewritten in the operation content setting unit 52, the control unit 5 causes the motor control unit 51 to update the operation commands for each motor and continues mowing the autonomous lawn mower 1.
[0051] The post-processing determination unit 54 determines whether post-processing of the lawn mowing is required based on the final processing state after the lawn mowing included in the reverse imaging result of the reverse imaging device 4. Specifically, the post-processing determination unit 54 determines whether post-processing is required based on the states of the mowed grass Gb after the lawn mowing, such as the appearance of unmowed grass, the appearance of fallen grass, and the color of the grass Gb. The abnormal notifications do not necessarily have to be limited to explicit notifications to external terminals, but may include information indicating that post-processing is required.
[0052] If the re-processing determination unit 54 determines that re-processing is required, the control unit 5 issues instructions for the re-processing. The re-processing instructions include a notification to the operations manager and a re-processing instruction for the autonomous lawn mower 1.
[0053] Post-processing is not limited to the operation performed by the autonomous lawn mower 1. Post-processing may also include notifications of post-processing content to external terminals such as mobile devices or personal computers belonging to farm managers via the communication unit 56. This allows, for example, requesting the farm manager to perform operations that cannot be performed by the autonomous lawn mower 1, such as removing clumps of grass clippings Gc1.
[0054] The positioning unit 55 measures the current position and direction of the autonomous lawn mower 1 currently mowing the lawn. In particular, any positioning unit 55 capable of measuring the current position of the autonomous lawn mower 1 while mowing the lawn can be used without restriction. For example, the positioning unit 55 can measure the current position of the autonomous lawn mower 1 using a GPS (Global Positioning System). The positioning unit 55 can use lasers or similar means to measure the distance and angle to specific locations, such as the charging station of the autonomous lawn mower 1, thereby determining the current position of the autonomous lawn mower 1.
[0055] The communication unit 56 enables wireless communication between the autonomous lawn mower 1 and external devices such as external terminals.
[0056] Back to Fig. 1: The storage unit 6 stores various data about the operation of the autonomous lawn mower 1 during lawn mowing. Specifically, the storage unit 6 stores the operation content of the autonomous lawn mower 1 during lawn mowing. The operation content includes information such as the forward imaging result acquired by the forward imaging device 3, the reverse imaging result acquired by the reverse imaging device 4, information about the position and imaged location of the autonomous lawn mower 1 measured by the positioning unit 55, and the operation content of the autonomous lawn mower 1.
[0057] The control unit 5, including the various functional blocks described above, consists of a processing unit such as a CPU (Central Processing Unit). To implement the functions of each functional block, the control unit 5 may include auxiliary storage devices such as HDD (Hard Disk Drive), which contain various control programs such as application software and OS (Operating System), as well as main storage devices such as RAM (Random Access Memory) for temporarily storing data required by the processing unit during program execution. The storage unit 6 may consist of storage devices such as RAM (Random Access Memory), for example.
[0058] The control unit 5 is not limited to integration into the lawnmower body 2. All or part of the functions of the control unit 5 can be provided, for example, in external terminals such as personal computers and connected via network lines, thereby enabling communication via network connections. Determination of the operating content of an autonomous lawnmower
[0059] Next, the control when determining the specific operation content of the autonomous lawn mower 1 in the present embodiment will be described. First, using Fig. 9, the movement path of the autonomous lawn mower 1 of the present embodiment during lawn mowing is described. Here, Fig. 9 shows a plan view of a rectangular work surface W. The autonomous lawn mower 1 is positioned at the start position for operation in the lower left corner of the work surface W before mowing the lawn.
[0060] During lawn mowing, the autonomous lawn mower 1 in the present embodiment moves linearly from the lower edge W1 to the upper edge W2 of the working area W, then changes direction by 180 degrees at the upper edge W2 and moves linearly toward the lower edge W1. Upon reaching the lower edge W1, the autonomous lawn mower 1 changes direction again by 180 degrees and moves linearly toward the upper edge W2. The autonomous lawn mower 1 repeats this movement between the edges W1 and W2 from the operation start position to the operation end position, mowing the lawn across the entire working area W. Information about such movement paths of the autonomous lawn mower 1 is prestored, for example, in the storage unit 6.
[0061] Fig. Figure 10 is a flowchart illustrating the control operation for determining the operation content of the autonomous lawn mower 1. The control unit 5 drives the traction motor 21, the blade motor 22, and the work lift motor 23 based on the operation content recorded in the operation content setting unit 52. As a result, the autonomous lawn mower 1 mows the lawn while moving along the movement path as shown in Fig. 9 shown.
[0062] When lawn mowing begins, the control unit 5 acquires the forward imaging result from the forward imaging device 3 and the reverse imaging result from the reverse imaging device 4 (step S11). As a result, the unmowed grass state Ga and the mowed grass state Gb are acquired. The control unit 5 performs reinforcement learning based on the grass states Ga and Gb (step S12).
[0063] Through the reinforcement learning by the operation content determining unit 53, the control unit 5 calculates rewards based on the state of grass Ga detected by the forward imaging result and the state of grass Gb detected by the backward imaging result, and determines a new operation content of lawn mowing, which is better than the current operation content, based on the rewards (step S13).
[0064] The information about the state of the unmowed grass Ga included in the forward imaging result preferably includes at least information about the moisture content of the grass Ga. This enables the control unit 5 to determine the operation content taking into account the moisture content of the grass Ga.
[0065] If the condition of the grass Ga included in the forward imaging result includes information about at least the number of grass Ga and / or the thickness of the grass Ga and information about the moisture content of the grass Ga, the control unit 5 preferably prioritizes the moisture content of the grass Ga over the number and thickness of the grass Ga when determining a new operation content. For example, even if the condition of the grass Ga is judged to be "normal," as in Fig. As shown in Figure 7, due to a small number of grass Ga or a small thickness of grass Ga, when the moisture content of grass Ga is high, the grass Ga may collapse under the weight of the moisture and tend to remain unmowed. In such cases, the condition of grass Ga is judged to be "poor." As a result, the control unit 5 can determine a more appropriate operation content based on the moisture content of the unmowed grass Ga included in the forward imaging result.
[0066] When the condition of the grass Ga included in the forward imaging result includes information about both the number and thickness of the grass Ga, the control unit 5 preferably prioritizes the thickness of the grass Ga over the number of grass Ga when determining a new operation content. When each individual grass Ga is thick, mowing is more difficult compared to mowing a plurality of thinner grasses and requires more torque for mowing. For example, even if the condition of the grass Ga is judged to be "normal" due to the presence of a plurality of thin grasses, as shown in Fig. 7, judged as "slightly poor" even when thick grass is present. As a result, the control unit 5 can determine a more appropriate operation content based on the thickness of the grass Ga.
[0067] If the grass condition Gb included in the reverse imaging result includes information about both the quantity of tufts of grass sections Gc1 and the condition of the grass Gb (e.g., unmowed grass, grass color), the control unit 5 preferably prioritizes the quantity of tufts of grass sections Gc1 over the condition of the grass Gb when determining a new operation content. The tufts of grass sections Gc1 may lead to grass death. For example, the condition of the grass Gb, as shown in Fig. 7, the condition of the grass Gb is judged as "good" due to the absence of uncut grass Gb or white tips Gb1, but when the amount of tufts of grass clippings Gc1 is high, the condition of the grass Gb is judged as "poor." As a result, the control unit 5 can determine a more appropriate operation content based on the amount of tufts of grass clippings Gc1.
[0068] When the condition of the grass Gb included in the reverse imaging result includes information about both the color of the grass Gb and the length unevenness of the grass Gb, the control unit 5 preferably prioritizes the information about the change from the color of the grass Ga included in the forward imaging result to the color of the grass Gb included in the reverse imaging result over the information about the length unevenness of the grass Gb when determining a new operation content. For example, even if the grass Gb is judged to be “good,” as in Fig. As shown in Figure 7, due to the absence of length unevenness, if the color of the grass Gb with the white tips Gb1 is whiter than the grass Ga, the condition of the grass Gb is judged to be "poor." As a result, the control unit 5 can determine a more appropriate operation content based on the appearance of the white tips Gb1 of the grass Gb.
[0069] One possible factor contributing to the presence of white tips Gb1 in the grass Gb may be the reduced sharpness of the worn cutting blades 241. Therefore, if the condition of the mown grass Gb is judged to be "poor" due to the presence of white tips Gb1, the control unit 5 may notify the farm manager's external terminal via the communication unit 56 and initiate the replacement of the cutting blades 241.
[0070] When the new operation content is determined, the control unit 5 overwrites the operation content recorded in the operation content setting unit 52 with the new operation content and causes the autonomous lawn mower 1 to operate based on the new operation content and continue mowing the lawn until the operation end position is reached (step S14). Post-processing by autonomous lawnmower
[0071] Next, the specific method for determining the necessity of post-processing by the autonomous lawn mower 1 in the present embodiment will be described using the flowchart of Fig. 11. Here, too, it is described how the autonomous lawnmower 1 mows the lawn along the path of movement in the working area W, similar to Fig. 9.
[0072] The autonomous lawn mower 1 performs the predetermined mowing of the vegetation within the work area W based on the operation content set in the operation content setting unit 52. While the autonomous lawn mower 1 is mowing the lawn, the control unit 5 determines whether post-processing is necessary based on the reverse imaging result acquired by the reverse imaging device 4. That is, when the lawn mowing starts, the control unit 5 acquires the reverse imaging result acquired by the reverse imaging device 4 at predetermined control intervals (step S21).
[0073] The control unit 5 determines whether post-processing is required based on the acquired reverse imaging result (step S22). Specifically, the control unit 5 causes the post-processing determination unit 54 to determine whether post-processing is required based on the final processing state of the mowed grass Gb included in the reverse imaging result.
[0074] If it is determined in step S22 that no post-processing is required (step S22; NO), the processing returns to step S21.
[0075] In the case where it is determined in step S22 that post-processing is required (step S22; YES), the control unit 5 issues an abnormal notification indicating that post-processing is necessary (step S23), and uses the abnormal notification as a trigger to store in the storage unit 6 information such as the imaged location from the reverse imaging result, which corresponds to the location for which a determination of post-processing was made (step S24).
[0076] Specifically, the storage unit 6 stores the lawn mowing operation content when it is determined that re-processing is required. The lawn mowing operation content includes information such as the imaged location (position of the autonomous lawn mower 1 as measured by the positioning unit 55) included in the reverse imaging result, information about the operation content during lawn mowing, and information about the re-processing content (the appearance of grass left unmowed, the appearance of fallen grass, etc.). The imaged location information is obtained from the position information measured by the imaging unit 55. Information about the traveling speed of the autonomous lawn mower 1 is obtained from the traveling commands output to the traveling motor 21.Information about the rotation speed of the cutting blades 241 is obtained from the travel commands issued to the blade motor 22. Information about the height of the cutting unit 24 is obtained from the travel commands issued to the unit lifting motor 23. The storage unit 6 stores the information about the travel speed of the autonomous lawn mower 1, the information about the rotation speed of the cutting blades 241, and the information about the height of the cutting unit 24 in conjunction with the information about the mapped location.
[0077] The control unit 5 repeats the processing of steps S21 to S24 until the autonomous lawn mower 1 reaches the operation end position. If a plurality of re-processing spots exist during lawn mowing, information about the operation content at each re-processing spot is stored in the storage unit 6.
[0078] The post-processing is carried out after the autonomous lawnmower 1 has finished mowing the lawn on the vegetation within the working area W. The sequence of the post-processing operation is described using the flowcharts in Fig. 12A and Fig. 12B. When mowing is finished, the autonomous lawnmower 1 stops driving the cutting blades 241.
[0079] When the autonomous lawn mower 1 reaches the operation end position and has completed mowing, the control unit 5 acquires information about the operation content at all post-processing spots stored in the storage unit 6 (step S31).
[0080] Next, the control unit 5 checks the operation content at each post-processing spot and instructs the post-processing of the vegetation at the imaged locations where post-processing is required based on the reverse imaging result. Specifically, the control unit 5 first determines whether there is any fallen grass (tufts of grass clippings Gc1) based on the operation content (step S32). A large amount of fallen grass may cause the grass to die, so the fallen grass must be removed immediately.
[0081] In step S32, as in Fig. 13, when one or more places with fallen grass (tufts of grass clippings Gc1) exist in the work area W after mowing the lawn (step S32; Yes), the control unit 5 obtains information about the amount of fallen grass at each place (amount of tufts of grass clippings Gc1) from the operation content (step S321).
[0082] The control unit 5 then compares the detected grass amount with a preset threshold (step S322). If the grass amount exceeds the threshold (step S322; Yes), the control unit 5 sends a notification to the manager's external terminal via the communication unit 56 and initiates follow-up processing (step S323). In other words, the control unit 5 notifies the manager that the amount of fallen grass exceeds the threshold at the specific locations on the work area W and requests the manager to remove the clumps of grass clippings Gc1.
[0083] After notifying the operator, the control unit 5 further determines whether any other re-treatment spots exist besides the re-treatment spots (step S324). Even if the amount of grass does not exceed the threshold in step S322 (step S322; No), the control unit 5 still determines whether other re-treatment spots exist (step S324). This is because a small amount of fallen grass is not considered a significant problem for the growth of the grass Gb and could instead be converted into soil fertilizer.
[0084] In the case of determining that no more post-processing spots exist in step S324 (step S324; No), the post-processing determination processing ends.
[0085] In the case of determining that another post-processing spot exists in step S324 (step S324; Yes), the control unit 5 checks whether a plurality of other post-processing spots exist (step S33).
[0086] If it is determined that only one more retouching spot exists in step S33 (step S33; No), the control unit 5 acquires the position information of this retouching spot from the operation content and instructs the autonomous lawn mower 1 to perform retouching (step S34). In other words, the control unit 5 causes the autonomous lawn mower 1 to travel directly to the retouching spot and perform retouching. As a result, the autonomous lawn mower 1 travels the shortest route from the operation end position to the retouching spot.
[0087] On the other hand, in a case of determining that a plurality of further post-processing spots exist in step S33 (step S33; Yes), the control unit 5 determines the sequence of post-processing for the plurality of post-processing spots and instructs the autonomous lawn mower 1 to perform the post-processing in the determined sequence (step S35).
[0088] The sequence of post-processing can be determined in order of proximity to the post-processing spots. For example, if, as in Fig. 13, two post-processing spots OB1 and OB2 are present in the work area W, the control unit 5 controls the autonomous lawn mower 1 to travel the shortest distance to each post-processing spot in order of proximity to the autonomous lawn mower 1, with the post-processing at each post-processing spot performed in a specific sequence, namely OB2 followed by OB1. This enables efficient post-processing and reduces unnecessary power consumption.
[0089] The sequence of post-processing can also be determined by the order of need for post-processing. As an example of the order of need for post-processing, post-processing spots with larger unmown grass areas can be prioritized. This allows for minimizing the poor appearance of the grass Gb in the working area W while taking remaining battery capacity into account.
[0090] Furthermore, the sequence of post-processing can be determined based on a comparison of the grass condition Gb, which includes the reverse imaging result for each post-processing spot. For example, post-processing is performed based on the order, such as the darkness of the grass color Gb, the number of uncut grass Gb, or the density of the uncut grass Gb, thereby ensuring effective post-processing tailored to the grass condition Gb in each post-processing spot.
[0091] Once the post-processing sequence is determined, the control unit 5 controls the autonomous lawn mower 1 to perform post-processing in the determined sequence. As a result, the autonomous lawn mower 1 performs post-processing at each post-processing spot (step S36).
[0092] When the autonomous lawn mower 1 performs re-mowing, it is preferable that the control unit 5 changes the work operation content during the re-mowing from the work operation content of the autonomous lawn mower 1 during normal operation (the first lawn mowing operation). Specifically, during the re-mowing, the control unit 5 changes at least the travel direction and travel speed of the autonomous lawn mower 1, the rotation speed of the cutting blades 241, and the height of the cutting unit 24. The lawn can be mowed under different operating conditions at re-mowing spots where normal operation has produced an unsatisfactory result, thereby enabling the operation efficiency to be improved or the finishing quality of the re-mowing to be further increased.
[0093] More specifically, the control unit 5 increases the travel speed of the autonomous lawn mower 1 for refinishing compared to the travel speed of the autonomous lawn mower 1 during normal operation. Since the density of the grass Gb at refinishing spots is already lower due to normal mowing operations, the control unit 5 can increase the travel speed of the autonomous lawn mower 1 during refinishing to improve efficiency without compromising the finishing quality.
[0094] When the travel speed of the autonomous lawn mower 1 is increased during post-cutting, the control unit 5 preferably increases the rotation speed of the cutting blades 241 for post-cutting compared to the rotation speed of the cutting blades 241 for normal operation. Increasing the rotation speed of the cutting blades 241 increases the frequency of the cutting blades 241 contacting the grass Gb per unit area, so that the grass Gb is unlikely to remain uncut even when the autonomous lawn mower 1 travels at higher speeds. Therefore, efficiency can be further improved.
[0095] When the autonomous lawnmower 1 performs a post-processing operation, the control unit 5 preferably changes the direction of travel of the autonomous lawnmower 1 for post-processing compared to the direction of travel for normal operation. For example, as shown in Fig. As shown in Figure 14, when the travel direction for normal operation of the autonomous lawn mower 1 is from bottom to top, the autonomous lawn mower 1 performs re-processing at the re-processing spot OB while traveling in a direction other than bottom to top. If the lawn is mowed for re-processing in the same travel direction as in normal operation, the conditions may be the same as in normal operation, resulting in the risk of the grass remaining unmowed. By deviating the travel direction of the autonomous lawn mower 1 during re-processing from the travel direction in normal operation, the quality of the re-processing can be further improved.
[0096] Fig. 14 illustrates the case where the autonomous lawnmower 1 performs post-processing while traveling in a direction orthogonal to the normal operation direction of travel. The direction of travel of the autonomous lawnmower 1 for post-processing is not limited to the direction orthogonal to the normal operation direction of travel, but may be any direction that intersects the normal operation direction of travel. The direction of travel of the autonomous lawnmower 1 for post-processing may be a direction opposite to the normal operation direction of travel.
[0097] In step S23 of the Fig. 11, an abnormal notification triggered by a determination that reworking is required may last for a predetermined time interval or a predetermined distance in the direction of travel of the autonomous lawnmower 1. Such abnormal notifications arise due to errors in the overlap width during lawn mowing. As shown in Fig. 15A, the lawn is typically mowed to create an overlap section 100 with a predetermined overlap width between two adjacent rows of work areas (work zones) in which the autonomous lawn mower 1 travels in directions that are 180 degrees different. However, if, as in Fig.15B, the two adjacent rows of work areas (work regions) do not overlap due to an error in the overlap width, a long unmown area 200 extends without overlap along the traveling direction of the autonomous lawn mower 1. In this case, the control unit 5 detects abnormalities over a predetermined time interval or a predetermined distance in the traveling direction of the autonomous lawn mower 1 and thus continues to issue abnormal notifications.
[0098] If the control unit 5 detects abnormal notifications in step S23 that continue for a predetermined time interval or a predetermined distance, the control unit 5 preferably causes the autonomous lawn mower 1 to perform follow-up processing while traveling along the normal operation travel direction. This enables efficient processing of the long unmowed area 200 caused by the overlap width error. If an abnormal notification in step S23 is resolved within a predetermined time interval or distance, the control unit 5 changes the travel direction of the autonomous lawn mower 1 during follow-up processing from the travel direction during normal operation, as described above.
[0099] While the autonomous lawn mower 1 travels to the refinishing spots, the cutting blades 241 remain stationary. For example, if there is only one refinishing spot, the cutting blades 241 begin to operate when the autonomous lawn mower 1 reaches the refinishing spot to begin refinishing. If there are multiple refinishing spots, the cutting blades 241 begin to operate when the autonomous lawn mower 1 reaches each refinishing spot to begin refinishing. In other words, after completing refinishing at one refinishing spot, the cutting blades 241 remain stationary until refinishing begins at the next refinishing spot. As a result, unnecessary power consumption can be reduced.
[0100] As soon as the post-processing at all post-processing spots is completed, the autonomous lawnmower 1 stops mowing the lawn in the working area W.
[0101] In the above embodiments, the control unit 5 instructs the re-processing of the fallen grass after the autonomous lawn mower 1 reaches the end position of operation and completes the lawn mowing, as in the case of instructing other re-processing operations; however, this is not limited to this. The control unit 5 may instruct the re-processing of the fallen grass when the fallen grass is detected.
[0102] In the above embodiments, the autonomous lawn mower 1 ejects the grass clippings Gc backward in the direction of travel; however, this is not limited to this. The autonomous lawn mower can also eject the grass clippings Gc transversely to the direction of travel. In this case, the autonomous lawn mower can additionally include an imager oriented laterally to the direction of travel to measure the amount of grass ejected laterally to the direction of travel (the appearance of the fallen grass). Explanation of reference symbols 1 autonomous lawnmower 2 lawn mower bodies 3 Forward imaging device (forward imager) 4 Reverse imaging device (reverse imager) 5 Control unit Ga unmown grass Gb mown grass Gc1 Tufts of grass clippings (grass thrown from the lawnmower body) QUOTES CONTAINED IN THE DESCRIPTION
[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature
[0000] JP 2019-88252
[0003]
Claims
[1] Autonomous lawnmower comprising: a lawnmower body that performs a lawnmowing operation during autonomous driving; a control unit that controls the lawnmower body based on a recorded operation content; a forward imager oriented forward in a direction of travel of the lawnmower body; and a reverse imager oriented backward in a direction of travel of the lawnmower body; wherein, when a reverse imaging result acquired by the reverse imager is within a predetermined range of conditions, the control unit determines the operation content based on a forward imaging result acquired by the forward imager. [2] The autonomous lawn mower according to claim 1, wherein the forward imaging result includes at least information of a state of uncut grass. [3] An autonomous lawn mower according to claim 2, wherein the information of a state of uncut grass includes at least information of a moisture content of grass. [4] The autonomous lawn mower according to claim 3, wherein the information of a state of uncut grass further includes information of at least one of a number of grasses and a thickness of grass, and the control unit prioritizes the information of a moisture content of grass over the information of the number of grasses and / or the thickness of grass when determining the operation content. [5] The autonomous lawn mower according to claim 4, wherein the information of a state of uncut grass further includes the information of both the number of grass and the thickness of grass, and the control unit prioritizes the information of a thickness of grass over the information of the number of grass when determining the operation content. [6] An autonomous lawn mower according to any one of claims 2 to 5, wherein the reverse imaging result includes at least information of an amount of grass discharged from the lawn mower body. [7] The autonomous lawn mower according to claim 6, wherein the reverse imaging result further includes information of a state of mowed grass, and the control unit prioritizes the information of the amount of grass over the information of a state of mowed grass when determining the operation content. [8] The autonomous lawn mower according to claim 7, wherein the information of a state of mowed grass includes information of a color of grass and a length unevenness of grass, and the control unit prioritizes information of a change of a color of unmowed grass included in the forward imaging result to a color of mowed grass included in the reverse imaging result over the information of a length unevenness of grass when determining the operation content.
Citation Information
Patent Citations
Mobile vehicle
JP2019088252A
2019-88252
