Control method and autonomous working machine

Abstract

The invention relates to a control method, a control device, a storage medium and an autonomous working machine. The control method comprises the following steps: acquiring a first map of a boundary of a working area; controlling the autonomous working machine to move according to the first map; in the moving process, feature supplementary collection is conducted on the first map, and the feature supplementary collection comprises the steps that in response to the fact that the autonomous working machine moves to all target positions, first updated images are collected at the target positions, and the target positions are located on the boundary or within the preset distance range from the boundary; the first updated image represents an image corresponding to the surrounding environment of the boundary; and updating the first map according to the first update image. According to the method and the device, the accuracy of the established first map can be improved, so that a favorable influence is generated on subsequent movement and / or work of the autonomous working machine.

Description

Technical Field

[0001] This invention relates to a control method, a control device, a storage medium, and an autonomous machine. Background Technology

[0002] As smart devices become increasingly common in daily life, autonomous machines (such as automatic lawnmowers for garden maintenance and robotic vacuum cleaners for home floor cleaning) are also gaining popularity among users.

[0003] Autonomous working machines typically move and / or operate within a designated work area. For example, an automatic lawnmower can travel across a user's lawn and perform mowing operations, achieving automated lawn cutting and significantly reducing labor costs.

[0004] Autonomous machines typically rely on pre-existing maps for movement and / or operation within their work area. This necessitates that the autonomous machine pre-map the work area, and the accuracy of the map significantly impacts its subsequent movement and / or operation. For example, in the case of automatic lawnmowers, map accuracy severely affects the cutting coverage rate during actual mowing operations. Summary of the Invention

[0005] In view of this, this application provides a control method, control device, storage medium, and autonomous working machine for an autonomous working machine, thereby improving the accuracy of the generated map.

[0006] In one embodiment, a control method for an autonomous working machine is provided, applied to the autonomous working machine, the control method comprising:

[0007] Get the first map of the work area boundaries;

[0008] Based on the first map, control the movement of the autonomous working machine;

[0009] During the movement, feature acquisition is performed on the first map. Feature acquisition includes: in response to the autonomous working machine moving to each target location, acquiring a first updated image at the target location, wherein the target location is located on the boundary or within a preset distance from the boundary, and the first updated image represents an image corresponding to the surrounding environment of the boundary; and updating the first map based on the first updated image.

[0010] According to some achievable methods in the embodiments of this application, during the movement process, an image of the work area (A) is acquired, and a pre-trained deep learning model is used to process the image of the work area (A) to identify the boundary between the lawn area and the non-lawn area, so as to control the autonomous working machine (10) to move within the work area (A).

[0011] According to some feasible methods in the embodiments of this application, controlling the movement of the autonomous working machine based on the first map includes:

[0012] Based on the first map, the autonomous working machine is controlled to move along the boundary.

[0013] According to some feasible methods in the embodiments of this application, controlling the movement of the autonomous working machine based on the first map includes:

[0014] Based on the first map, the autonomous working machine is controlled to move and operate within the working area.

[0015] According to some achievable methods in the embodiments of this application, the feature acquisition is terminated in response to the autonomous working machine returning to the starting position of the movement, or the distance between the machine position of the autonomous working machine and the starting position being less than a distance threshold.

[0016] According to some implementable methods in the embodiments of this application, the method further includes:

[0017] Determine the boundary segment traversed by the feature acquisition;

[0018] In response to the formation of a first sub-region in the boundary segment, the autonomous working machine is controlled to operate in the first sub-region;

[0019] In response to the completion of the work of the autonomous machine in the first sub-region, the process returns to executing according to the first map, controlling the autonomous machine to move along the boundary until the autonomous machine returns to the starting position of the movement.

[0020] According to some implementable methods in the embodiments of this application, the first map includes boundary information corresponding to each of multiple boundary segments;

[0021] Controlling the movement of the autonomous working machine according to the first map includes: determining a second sub-region based on the boundary information corresponding to multiple boundary segments included in the first map; and controlling the autonomous working machine to move along the boundary segments of the second sub-region.

[0022] The control method further includes: in response to completing feature acquisition of the boundary segment of the second sub-region, controlling the autonomous working machine to move and work in the second sub-region.

[0023] According to some achievable methods in the embodiments of this application, the first map includes boundary information corresponding to unclear boundary segments, wherein when the autonomous working machine is located in the unclear boundary segment, the autonomous working machine cannot identify the boundary position based solely on the image;

[0024] Controlling the movement of the autonomous working machine according to the first map includes:

[0025] Based on the first map, the autonomous working machine is controlled to move along the unclear boundary section.

[0026] According to some achievable methods in the embodiments of this application, controlling the autonomous working machine to move along the unclear boundary segment based on the first map includes: starting from a boundary position where the positioning accuracy meets the visual loop closure condition, controlling the autonomous working machine to move along the unclear boundary segment based on the first map.

[0027] According to some feasible methods in the embodiments of this application, controlling the autonomous working machine to move and work within the working area based on the first map includes:

[0028] Based on the planned movement mode, the autonomous working machine is controlled to move and work within the work area, which may be a part or all of the work area.

[0029] According to some achievable methods in the embodiments of this application, the autonomous working machine performs feature supplementation on the first map in at least some rounds of work tasks, and the working path of the autonomous working machine (10) is different in work tasks in which feature supplementation is performed and in work tasks in which feature supplementation is not performed.

[0030] According to some achievable methods in the embodiments of this application, the difference in the working path of the autonomous working machine (10) in a task where feature acquisition is performed and in a task where feature acquisition is not performed includes at least one of the following:

[0031] The path spacing of the work path is different in work tasks that perform feature acquisition and those that do not.

[0032] The path direction of the work path is different in work tasks where feature supplementation is performed and in work tasks where feature supplementation is not performed.

[0033] The autonomous machine (10) travels at different speeds along the work path during a task involving feature acquisition and during a task where feature acquisition is not performed.

[0034] According to some achievable methods in the embodiments of this application, the working path of the autonomous working machine is different in different rounds of global work tasks. The global work task is based on a planned movement mode, which controls the autonomous working machine to move and work within the work area.

[0035] According to some achievable methods in the embodiments of this application, the first map includes boundary information corresponding to clear boundary segments, and when the autonomous working machine is located in the clear boundary segment, the autonomous working machine can identify the boundary position based solely on the image.

[0036] Based on a planned movement pattern, the autonomous working machine is controlled to move and operate within the work area, including:

[0037] Based on the boundary information corresponding to the clear boundary segment, a third sub-region is determined in the working area, and at least one edge of the third sub-region is the clear boundary segment;

[0038] Based on the planned movement pattern, the autonomous working machine is controlled to move and operate within the third sub-region.

[0039] According to some achievable methods in the embodiments of this application, the first map includes boundary information corresponding to clear boundary segments, and when the autonomous working machine is located in the clear boundary segment, the autonomous working machine can identify the boundary position based solely on the image.

[0040] The target location is located on or within a preset distance from the clear boundary segment.

[0041] According to some feasible methods in the embodiments of this application, controlling the autonomous working machine to move and work within the working area based on the first map includes:

[0042] Based on the first map, at least two target boundary segments are determined, and the autonomous working machine is controlled to move along the target boundary segments;

[0043] A semi-random movement mode is used between the target boundary segments to control the autonomous working machine to move from one target boundary segment to another by traversing the interior of the working area.

[0044] According to some achievable methods in the embodiments of this application, the target boundary segment is a clear boundary segment, and when the autonomous working machine is located in the clear boundary segment, the autonomous working machine can identify the boundary position based solely on the image.

[0045] Alternatively, the target boundary segment is an unclear boundary segment, and when the autonomous machine is located in the unclear boundary segment, the autonomous machine cannot identify the boundary position based solely on the image.

[0046] Alternatively, the target boundary segment is part of the boundary of the working area, and the positioning accuracy of the starting position in the target boundary segment meets the visual loop closure condition;

[0047] Alternatively, the target boundary segment is part of the boundary of the working area, and the starting position of the target boundary segment is located in the clear boundary segment.

[0048] According to some feasible methods in the embodiments of this application, using a semi-random movement mode to control the autonomous working machine to move from one target boundary segment to another through the interior of the working area includes:

[0049] In the target boundary segment L2, a reference position is determined. The distance between the reference position and the previous target position in the other target boundary segment is greater than a first threshold. Alternatively, the reference position is the starting position of the other target boundary segment. Alternatively, the reference position is the previous target position in the other target boundary segment. Alternatively, the reference position is located between adjacent historical target positions.

[0050] Based on a semi-random movement mode, the autonomous working machine is controlled to pass through the interior of the working area and move to the reference position.

[0051] According to some achievable methods in the embodiments of this application, if the reference position is located between adjacent historical target positions, then during the process of controlling the autonomous working machine to pass through the interior of the working area based on a semi-random movement mode, in response to the autonomous working machine detecting a prohibited area, the autonomous working machine is controlled to stop moving to the reference position and return to the target boundary segment.

[0052] According to some feasible methods in the embodiments of this application, controlling the autonomous working machine to move and work within the working area based on the first map includes:

[0053] Based on the first map, multiple reversal points are determined on the boundary of the work area;

[0054] The autonomous working machine is controlled to move and work within the working area in a semi-random movement mode. In the semi-random movement mode, the autonomous working machine moves to the reversing position, turns at the reversing position, and moves to other reversing positions.

[0055] According to some achievable methods in the embodiments of this application, the autonomous working machine moving to the target location includes:

[0056] During the movement, the machine position of the autonomous working machine and the first information corresponding to the machine position are determined. The first information includes at least one of the following parameters: visual positioning accuracy, distance between the machine position and the previous target position, and time interval between the current time point and the time point when the machine moves to the previous target position.

[0057] In response to the first information satisfying the first condition, it is determined that the autonomous working machine has moved to the target location.

[0058] According to some achievable methods in the embodiments of this application, the first map includes boundary information corresponding to unclear boundary segments, wherein when the autonomous working machine is located in the unclear boundary segment, the autonomous working machine cannot identify the boundary position based solely on the image;

[0059] In response to the first information satisfying the first condition, determining that the autonomous working machine has moved to the target location includes:

[0060] In response to the first information satisfying the first condition and the machine location being within the unclear boundary segment, it is determined that the autonomous working machine moves to the target location.

[0061] According to some implementable methods in the embodiments of this application, in response to the first information satisfying the first condition, determining that the autonomous working machine has moved to the target location includes:

[0062] In response to the machine location being located on or near the boundary, and the positioning accuracy of the machine location meeting the visual loop closure condition, the machine location is determined as the target location.

[0063] According to some feasible methods in the embodiments of this application, acquiring the first updated image at the target location includes:

[0064] At the target location where the autonomous machine turns based on the first map, the first updated image is captured by the camera of the autonomous machine.

[0065] According to some feasible methods in the embodiments of this application, a first updated image is acquired at the target location, including:

[0066] The autonomous working machine is controlled to rotate at the target position, and during the rotation, the first updated image is acquired based on the camera of the autonomous working machine. The first updated image includes at least two images from different angles.

[0067] Alternatively, the camera can be controlled to rotate relative to the body of the autonomous working machine, and during the rotation of the camera, the first updated image can be acquired based on the camera of the autonomous working machine, wherein the first updated image includes at least two images from different angles;

[0068] Alternatively, the panoramic camera of the autonomous working machine can be controlled to capture panoramic images, and the first updated image is the panoramic image.

[0069] According to some feasible methods in the embodiments of this application, acquiring the first updated image at the target location includes:

[0070] At the target location where the autonomous working machine (10) turns based on the first map, the first updated image is acquired by the camera of the autonomous working machine (10).

[0071] According to some feasible methods in the embodiments of this application, at the target location where the autonomous working machine (10) turns based on the first map, the first updated image is acquired through the camera of the autonomous working machine (10), including:

[0072] When the autonomous working machine (10) turns in place along the planned path based on the first map, it acquires the first updated image through the camera of the autonomous working machine (10).

[0073] According to some achievable methods in the embodiments of this application, the in-situ turning includes:

[0074] Control the autonomous working machine (10) to stop moving forward;

[0075] The autonomous working machine (10) is controlled to rotate around a preset rotation center by a preset angle, wherein the preset angle is less than or equal to 180 degrees.

[0076] According to some implementable methods in the embodiments of this application, obtaining the first map of the boundary of the working area includes:

[0077] The autonomous working machine is controlled to move along the boundary and perform mapping. The mapping includes: acquiring inertial navigation information from the inertial navigation unit of the autonomous working machine and a first initial image captured by the camera of the autonomous working machine, wherein the shooting angle corresponding to the first initial image is different from the shooting angle corresponding to the first updated image; generating the first map using the inertial navigation information and visual features extracted from the first initial image; or,

[0078] Receive the first map from the server or user terminal; or,

[0079] Receive a map download instruction from the server or user terminal, and download the first map according to the map download instruction.

[0080] According to some feasible methods in the embodiments of this application, the movement direction of the autonomous working machine is the same or opposite during the mapping process and the feature acquisition process.

[0081] According to some achievable methods in the embodiments of this application, the mapping further includes: acquiring a second updated image at the mapping start position on the boundary, wherein the second updated image represents an image corresponding to the surrounding environment of the mapping start position;

[0082] In response to the autonomous machine returning to the mapping start position, the mapping process ends.

[0083] According to some implementable methods in the embodiments of this application, updating the first map based on the first updated image includes:

[0084] Obtain the inertial navigation information corresponding to the first updated image;

[0085] Determine the feature points of the first updated image;

[0086] Determine the descriptor information of the feature point, the position information of the feature point in the image, and the three-dimensional coordinate information of the feature point;

[0087] The first map is updated based on the inertial navigation information, the descriptor information of the feature points, the position information of the feature points in the image, and the three-dimensional coordinate information of the feature points.

[0088] According to some implementable methods in the embodiments of this application, the control method further includes:

[0089] A second map is generated using the first map. The second map includes coordinate information and visual information of multiple locations on the boundary, and coordinate information and visual information of multiple locations within a preset distance from the boundary.

[0090] The second map is sent to the user terminal so that the user terminal displays the second map.

[0091] According to some implementable methods in the embodiments of this application, the control method further includes:

[0092] Receive boundary attribute information sent by the user terminal;

[0093] Based on the boundary attribute information, the clear boundary segments and unclear boundary segments in the first map are determined.

[0094] According to some feasible methods in the embodiments of this application, the autonomous working machine is controlled to work before it moves according to the first map and performs feature supplementation.

[0095] Alternatively, while controlling the autonomous working machine to move and perform feature acquisition based on the first map, the autonomous working machine can be controlled to work.

[0096] Alternatively, after controlling the autonomous working machine to move and perform feature acquisition based on the first map, the autonomous working machine can be controlled to work.

[0097] The control of the autonomous working machine to move according to the first map includes controlling the autonomous working machine to move along the boundary according to the first map, or controlling the autonomous working machine to move within the working area according to the first map.

[0098] In a second aspect, a control device is provided for an autonomous working machine configured to move and / or operate in a work area, the autonomous working machine being equipped with a camera, the control device comprising:

[0099] The map acquisition unit is configured to acquire a first map of the boundaries of the work area;

[0100] A motion control unit is configured to control the movement of the autonomous working machine according to the first map;

[0101] A feature acquisition unit is configured to acquire features of the first map during the movement. The feature acquisition includes: in response to the autonomous working machine moving to each target location, acquiring a first updated image at the target location, wherein the target location is located on the boundary or within a preset distance from the boundary, and the first updated image represents an image corresponding to the surrounding environment of the boundary; and updating the first map based on the first updated image.

[0102] Thirdly, a computer-readable storage medium is provided, the storage medium storing a computer program for performing the above-described methods.

[0103] Fourthly, an autonomous working machine is provided, comprising:

[0104] processor;

[0105] Memory used to store the processor's executable instructions;

[0106] The processor is used to execute the above method.

[0107] As can be seen from the above technical solutions, after obtaining the first map of the boundary of the work area, this application controls the autonomous working machine to perform feature supplementation on the first map during its movement. That is, it collects the first updated image at the target position on the boundary or within a preset distance from the boundary and uses the first updated image to update the first map, thereby improving the accuracy of the established first map and thus having a beneficial impact on the subsequent movement and / or work of the autonomous working machine. Attached Figure Description

[0108] The objectives, technical solutions, and beneficial effects of the present invention described above can be clearly obtained through the following detailed description of specific embodiments that enable the implementation of the present invention, in conjunction with the accompanying drawings.

[0109] The same reference numerals and symbols in the accompanying drawings and the specification are used to represent the same or equivalent elements.

[0110] Figure 1 A schematic structural diagram of an autonomous mobile machine provided for an exemplary embodiment of this application;

[0111] Figure 2 Another schematic structural diagram of the autonomous working machine provided in the embodiments of this application;

[0112] Figure 3 A flowchart illustrating the control method for an autonomous working machine provided in this application embodiment;

[0113] Figure 4a A schematic diagram of the feature acquisition path provided in Embodiment 1 of this application;

[0114] Figure 4b A schematic diagram of the working path provided in Embodiment 1 of this application;

[0115] Figure 5a A schematic diagram of the path and working path for supplementary sampling of one of the features provided in Embodiment 2 of this application;

[0116] Figure 5b A schematic diagram of the path and working path for another feature supplementary sampling provided in Embodiment 2 of this application;

[0117] Figure 6a A schematic diagram of the path and working path for supplementary sampling of one of the features provided in Embodiment 3 of this application;

[0118] Figure 6b A schematic diagram of the path and working path for another feature supplementary sampling provided in Embodiment 3 of this application;

[0119] Figure 7a A schematic diagram of the path for supplementary sampling of one of the features provided in Embodiment 4 of this application;

[0120] Figure 7b A schematic diagram of a path for simultaneous feature acquisition during operation, provided for Embodiment 4 of this application;

[0121] Figure 7c This is a schematic diagram of another path for simultaneous feature acquisition during operation, provided in Embodiment 4 of this application.

[0122] Figure 8aA schematic diagram of a working path provided for Embodiment 5 of this application;

[0123] Figure 8b A schematic diagram of the feature acquisition path and working path provided for Embodiment 5 of this application;

[0124] Figure 9 A schematic diagram of the path for simultaneous feature acquisition during operation, provided in Embodiment Six of this application;

[0125] Figure 10 A schematic diagram of the path for simultaneous feature acquisition during operation, provided in Embodiment 7 of this application;

[0126] Figure 11 A schematic diagram of the path for simultaneous feature acquisition during operation, provided in Embodiment 8 of this application;

[0127] Figure 12a A schematic diagram of the path for simultaneous feature acquisition during operation, provided for Embodiment 9 of this application;

[0128] Figure 12b A schematic diagram illustrating the selection of the target boundary segment provided in Embodiment 9 of this application;

[0129] Figure 12c This is a schematic diagram illustrating the handling of prohibited areas during the process of simultaneous feature acquisition and operation, as provided in Embodiment 9 of this application.

[0130] Figure 13a A schematic diagram of the path for the first feature acquisition provided in Embodiment 10 of this application;

[0131] Figure 13b A schematic diagram of the path for the second feature acquisition provided in Embodiment 10 of this application;

[0132] Figure 14 A schematic block diagram of a control device provided in an embodiment of this application;

[0133] Figure 15 A schematic block diagram of an autonomous working machine provided in an embodiment of this application. Detailed Implementation

[0134] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention. The embodiments provided in this specification can be combined with each other.

[0135] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0136] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0137] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0138] The embodiments provide an autonomous working machine, a control method, and a computer-readable storage medium. The autonomous working machine can be an intelligent device capable of automatic movement, such as an automatic lawnmower, automatic vacuum cleaner, automatic mop, or automatic snowplow. It can automatically move and perform corresponding tasks within a designated working area, and can also return to a docking station along the boundary of the working area for parking or charging.

[0139] This embodiment provides an autonomous working machine 10, such as Figure 1 and Figure 2 As shown, the autonomous working machine 10 includes a body 100, an imaging sensor 200, a position sensor 500, and a control circuit 600.

[0140] Specifically, the autonomous working machine 10 includes a drive unit 700 disposed on the machine body 100. The drive unit 700 is used to move the machine body 100 on the working surface according to the received drive commands. It typically includes rollers and a motor that drives the rollers to rotate. The rollers may include driving rollers and driven rollers. The rollers may be distributed on both sides of the machine body 100, and the number of rollers on each side may be one or two, etc.

[0141] The autonomous working machine 10 also includes a working module, which is used to perform specific work tasks. For example, if the autonomous working machine 10 is an automatic lawnmower, the working module includes lawnmower blades, a cutting motor, etc., and may also include auxiliary components such as a lawnmower height adjustment mechanism to optimize or adjust the lawnmower effect; if the autonomous working machine 10 is an automatic vacuum cleaner, the working module includes working components such as a vacuum motor, a vacuum port, a vacuum hose, a vacuum chamber, and a dust collection device for performing vacuuming tasks.

[0142] The autonomous working machine 10 may also include an energy module for providing energy for various tasks of the autonomous working machine 10. The energy module may include a rechargeable battery and a charging connection structure, wherein the charging connection structure is typically a charging electrode plate that can be used in conjunction with a charging electrode plate provided at the docking station to charge the autonomous working machine 10.

[0143] The autonomous working machine 10 also includes a memory 400 for storing data generated by sensors or control circuits, or pre-storing data for use by the control circuits.

[0144] The autonomous working machine 10 also includes a position sensor 500, which may include an IMU (inertial measurement unit) or an ODO (odometer) mounted on the drive unit 700, for obtaining the relative position based on the movement of the body 100.

[0145] In addition to the modules mentioned above, the autonomous working machine 10 may also include a housing for accommodating and installing the various modules, a control panel for user operation, and various environmental sensors, such as humidity sensors, temperature sensors, acceleration sensors, and light sensors. These sensors can assist the autonomous working machine 10 in determining the working environment in order to execute the corresponding program.

[0146] The control circuit 600 is the core component of the autonomous working machine 10. It is used to control the autonomous working machine 10 to move and work automatically. Its functions include controlling the working module to start or stop, controlling the drive device 700 to move, judging the power of the energy module and controlling the autonomous working machine 10 to return to the docking station for automatic docking and charging, and executing corresponding programs based on the data from environmental sensors.

[0147] Reference Figure 1 and Figure 2 The autonomous working machine 10 includes an imaging sensor 200 connected to the body 100 for acquiring images along the forward direction of the body 100. These images are at least partially images of the working surface along the forward direction. The acquired images are located within the field of view 210 of the imaging sensor 200. The imaging sensor 200 can be a commonly used camera or lidar, etc.

[0148] Generally, the imaging sensor 200 is mounted on the upper front part of the fuselage 100, preferably centered, with its viewing angle pointing downwards and forwards to capture images of the working surface. The size of its field of view 210 can be adjusted according to actual needs; a larger field of view 210 captures more images in the forward direction of the fuselage 100, and vice versa. The forward direction of the fuselage 100 can be various, such as normal forward movement, backward movement, or turning. In this embodiment, the forward direction of the fuselage 100 refers to the normal forward movement direction, i.e., the direction of the central axis of the fuselage 100.

[0149] To address the technical problem of improving map accuracy, thereby enhancing machine positioning precision and coverage, and ultimately improving user satisfaction, this embodiment provides a control method applied to an autonomous working machine 10. The autonomous working machine 10 is configured to move and / or work in a work area. The autonomous working machine 10 is equipped with a camera (imaging sensor 200), which can be located on the front side of the machine body and can capture images of the work area.

[0150] Figure 3 This is a flowchart illustrating a control method for an autonomous working machine 10 provided in some embodiments of this application. The method is applied to the autonomous working machine 10, such as... Figure 3 As shown, the method mainly includes the following steps:

[0151] Step 301: Obtain the first map of the work area boundaries;

[0152] Step 303: Based on the first map, control the autonomous working machine 10 to move;

[0153] Step 305: During the movement, feature acquisition is performed on the first map. Feature acquisition includes: in response to the autonomous working machine 10 moving to each target location, acquiring a first updated image at the target location. The target location is located on the boundary or within a preset distance from the boundary. The first updated image represents the image corresponding to the surrounding environment of the boundary. The first map is updated based on the first updated image.

[0154] As can be seen, the method provided in some embodiments of this application, after obtaining a first map of the boundary of the working area, controls the autonomous working machine 10 to perform feature supplementation on the first map based on VSLAM technology during its movement. That is, it collects a first updated image at the target location on or within a certain range of the boundary and uses the first updated image to update the first map, thereby improving the accuracy of the established first map and thus having a beneficial impact on the subsequent movement and / or work of the autonomous working machine 10. The location needs to meet certain conditions to be considered a target location. These conditions include, for example, the location being within 0.2 meters of the boundary, or within 0.1 meters of the boundary, or the location being a location where a visual loop occurs on the boundary, or the location being a location where a visual loop occurs near the boundary, etc.

[0155] In some embodiments of this application, the acquired first map includes the boundary of the work area. The type of work area can be determined based on the type of autonomous machine 10. For example, if the autonomous machine 10 is a lawnmower, the work area is actually a lawn (also called a meadow, etc.). As another example, if the autonomous machine 10 is a snowplow, the work area is actually an area covered by snow. As yet another example, if the autonomous machine 10 is a robotic vacuum cleaner, the work area is actually a cleaning area.

[0156] The boundary of the work area is mainly used to distinguish between the work area and the non-work area. In some embodiments of this disclosure, the boundary of the work area may include one or more of the following: the outer boundary of the work area, the boundary of the passage within the work area, the boundary of the obstacle within the work area, etc.

[0157] In some embodiments of this application, the methods for obtaining the first map may include, but are not limited to, the following:

[0158] The first method: control the autonomous working machine 10 to move along the boundary of the working area and build a map to obtain the first map.

[0159] The mapping process described above may include: acquiring inertial navigation information from the inertial navigation unit of the autonomous machine 10 and a first initial image captured by the camera of the autonomous machine 10; and generating a first map based on VSLAM technology, using the inertial navigation information and visual features extracted from the first initial image. Typically, the shooting angle corresponding to the first initial image differs from the shooting angle corresponding to the first updated image involved in feature acquisition in step 305 of this embodiment.

[0160] For example, the autonomous machine 10 can move along the boundary of the work area under user control, or the autonomous machine 10 can automatically identify the boundary of the work area using AI (Artificial Intelligence) based on images captured by a camera, and then move along the boundary of the work area. During the movement, inertial navigation information from the inertial navigation unit of the autonomous machine 10 and the first initial image captured by the camera of the autonomous machine 10 are acquired.

[0161] The inertial navigation unit (IMU) includes a gyroscope and an accelerometer. The gyroscope is mainly used to measure angular acceleration, while the accelerometer is used to measure the acceleration of an object's motion. Given initial conditions, the inertial navigation module can determine its current position, yaw angle (mainly reflecting direction), and velocity without the need for external references, primarily reflecting the pose of the autonomous machine 10.

[0162] After obtaining inertial navigation information and the first initial image, visual features can be extracted from the first initial image based on VSLAM technology, and the first map can be obtained using the inertial navigation information and visual features.

[0163] In some embodiments of this application, the visual features extracted from the image may include, but are not limited to, one or any combination of: feature points, descriptor information of feature points, position information of feature points in the image, three-dimensional coordinate information of feature points, edges, textures, histograms, etc.

[0164] During the mapping process, the autonomous machine 10 moves smoothly along the boundary, and the mapping time is short, completing the mapping quickly. Because of the short time, the number of initial images is small, resulting in fewer visual features in the first map.

[0165] Typically, at least one image in the first updated image corresponds to a shooting angle that is different from the shooting angle corresponding to the first initial image. Different shooting angles can capture different objects, thereby obtaining more other visual features from the first updated image. Then, in step 305, more visual features are used to update the first map, enriching the visual features of the first map.

[0166] For example, the first map can be updated based on the descriptor information of the feature points corresponding to the first updated image, the location information of the feature points in the image, and the three-dimensional coordinate information of the feature points.

[0167] Furthermore, during the above-mentioned mapping process (the "mapping process" referred to in this application embodiment refers to the process of establishing a working map for the first time), a second updated image can be acquired at the mapping start position on the boundary. The second updated image represents the image corresponding to the surrounding environment at the mapping start position. In response to the autonomous working machine 10 returning to the mapping start position, the above-mentioned mapping process ends.

[0168] In other words, the mapping process described above can begin from a certain position on the boundary (i.e., the mapping start position). When starting mapping from the mapping start position, the second updated image is first acquired, and then the autonomous working machine 10 is controlled to move along the boundary of the working area and build the map until it returns to the mapping start position after circling the boundary of the working area, at which point the mapping ends.

[0169] Based on the images captured in real time by the camera and the second updated image captured at the beginning of the mapping, if a visual loop closure can be successfully performed, it means that the autonomous working machine 10 returns to the beginning of the mapping along the boundary of the working area, forming a closed boundary trajectory, and the mapping ends.

[0170] Acquiring the second updated image when starting mapping at the initial mapping position allows for the acquisition of more visual features near the initial mapping position, which can improve the success rate of visual loopback when returning to the initial mapping position along the edge, thereby improving mapping accuracy.

[0171] The second method: Receive the first map from the server or user terminal.

[0172] This method actually involves the autonomous working machine 10 obtaining a first map that has already been established from the server or user terminal. This first map can be one that was previously established by the autonomous working machine 10, or it can be one that was established by other autonomous working machines 10 and then sent to the server or user terminal for storage.

[0173] The content of the first map is the same as that in the first method described above, and will not be repeated here.

[0174] The third method: Receive map download instructions from the server or user terminal, and download the first map according to the map download instructions.

[0175] In this method, the autonomous worker machine 10 downloads a pre-established first map in response to a map download command. This map download command can be sent to the autonomous worker machine 10 by the server or by the user terminal. The autonomous worker machine 10 can download the first map from a specified address according to the map download command; this specified address can point to a server, user terminal, storage device, etc. Furthermore, the first map can be one previously established by the autonomous worker machine 10 or one established by another autonomous worker machine 10.

[0176] The content of the first map is the same as that in the first method described above, and will not be repeated here.

[0177] For the autonomous machine 10, boundary-related information on the map is crucial, as the accuracy of boundary-related information affects the positioning accuracy, movement accuracy, and safety of the machine's movement. The first map acquired in step 301 may have insufficient or inaccurate visual features. Therefore, while controlling the autonomous machine 10's movement based on the first map, it is necessary to acquire a first updated image. This updated image supplements the first map with visual features of the surrounding environment, enriching its visual features and improving its accuracy. This enhances the positioning and movement accuracy of the autonomous machine 10 when moving / working based on VSLAM technology and the first map, preventing it from going out of bounds and improving safety.

[0178] In some embodiments, the control method for the autonomous working machine 10 described above further includes the following steps:

[0179] Step 307 includes: during the movement, acquiring an image of the work area A, processing the image of the work area A using a pre-trained deep learning model, identifying the boundary between the lawn area and the non-lawn area, so as to control the autonomous working machine 10 to move within the work area A.

[0180] Taking an autonomous working machine as an example, as the automatic lawnmower moves according to the first map, the automatic lawnmower continuously or intermittently acquires environmental images of its front or surrounding working area (A) through its onboard camera (which may also be combined with near-infrared and / or depth cameras).

[0181] The autonomous machine in this example also includes a controller that communicates with the image acquisition module (camera), which integrates or externally connects to an image processing module. At the core of this image processing module is a pre-trained deep learning model (such as U-Net, a lightweight variant of DeepLab). This model is configured to perform pixel-by-pixel classification of the acquired environmental images, distinguishing between "grass" and "non-grass." After post-processing steps such as boundary extraction, smoothing, and setting a safety buffer, accurate boundary information is provided to the automatic lawnmower's navigation system, guiding the automatic lawnmower to avoid crossing the boundary between grass and non-grass areas, allowing it to work efficiently within a safe area (the grassy area).

[0182] In some embodiments, step 303 includes: controlling the autonomous working machine 10 to move along the boundary according to the first map.

[0183] After obtaining the first map, the autonomous working machine 10 moves and / or works along the boundary according to the first map. While the autonomous working machine 10 moves along the boundary, feature acquisition is performed, which allows for more targeted feature acquisition along the boundary and high acquisition efficiency. In some embodiments, feature acquisition ends in response to the autonomous working machine 10 returning to the starting position of the movement, or the distance between the machine position of the autonomous working machine 10 and the starting position being less than a distance threshold.

[0184] Starting from the initial position, the autonomous working machine 10 moves and / or works along the boundary according to the first map, while simultaneously acquiring additional features, until it returns to or near the initial position. In this way, the autonomous working machine 10 completes a full circle along the boundary, fully and completely adding the visual features of the entire boundary to the first map. The initial position can be the mapping starting point or any position on the boundary. The autonomous working machine 10 can use VSLAM technology for localization to determine whether its current position is the initial position. The autonomous working machine 10 moves along the boundary while simultaneously acquiring additional features; for example, the specific process can be found in Embodiment 1 below.

[0185] In some embodiments, the autonomous working machine 10 moves in the same direction during the mapping process and the feature acquisition process.

[0186] For example, during the mapping process, the autonomous working machine 10 moves counterclockwise along the boundary, and during the feature acquisition process, the autonomous working machine 10 also moves counterclockwise along the boundary.

[0187] In some embodiments, the autonomous working machine 10 moves in opposite directions during the mapping process and the feature acquisition process.

[0188] For example, during the mapping process, the machine moves at least one circle along the boundary in a counterclockwise direction, and during the feature acquisition process, the machine moves at least one circle along the boundary in a clockwise direction; or, during the mapping process, the machine moves at least one circle along the boundary in a clockwise direction, and during the feature acquisition process, the machine moves at least one circle along the boundary in a counterclockwise direction.

[0189] During feature acquisition, the machine acquires the first updated image from a different perspective than during the mapping process, thereby obtaining more visual features. Moving directly along the boundary in the opposite direction to the mapping process while simultaneously acquiring new features is a more efficient method.

[0190] In some embodiments, in step 303, the autonomous working machine 10 does not need to move a full circle along the boundary; it can move and / or work only along a portion of the boundary while simultaneously performing feature acquisition. For example, it can move along unclear boundary segments and perform feature acquisition, thereby improving the efficiency of feature acquisition.

[0191] The autonomous working machine 10 moves along a partial boundary and performs feature acquisition. The partial boundary can be a clear boundary segment or an unclear boundary segment, or a boundary segment specified by the user, etc. For example, the specific process can be found in Embodiments 2 to 5 below.

[0192] In some embodiments, step 303 includes: controlling the autonomous working machine 10 to move and work within the work area according to the first map.

[0193] After obtaining the first map, the autonomous working machine 10 moves and works within the working area based on the first map. While the autonomous working machine 10 is working within the working area, it performs feature supplementation, which can take into account both feature supplementation at the boundary and working within the area. There is no need to control the autonomous working machine 10 to perform feature supplementation separately. Users can perceive that the machine enters the area to work in a timely manner after mapping, which can demonstrate the timeliness of the machine's work within the area and the machine's intelligence to users.

[0194] In some embodiments, controlling the autonomous working machine 10 to move and work within a working area according to a first map includes: controlling the autonomous working machine 10 to move and work within a work area based on a planned movement pattern, wherein the work area is a part or all of the working area.

[0195] The planned movement mode can be a bow-shaped movement path, in which the autonomous worker 10 moves in a bow shape and performs tasks within the work area. The path in the planned movement mode can also be a zigzag path, etc. The planned movement mode can improve the work coverage of the autonomous worker 10 within the work area.

[0196] The autonomous working machine 10 moves within the working area based on a planned movement pattern and performs feature acquisition; for example, the specific process can be found in Embodiments 2 to 6 below. Furthermore, it can be combined with a method of moving along a boundary and performing feature acquisition.

[0197] In some embodiments, controlling the autonomous working machine 10 to move and work within a working area according to a first map includes: determining at least two target boundary segments according to the first map, and controlling the autonomous working machine 10 to move along the target boundary segments;

[0198] In this process, a semi-random movement mode is used between target boundary segments, controlling the autonomous working machine 10 to move from one target boundary segment to another by traversing the interior of the working area.

[0199] In semi-random movement mode, the machine turns when it encounters a boundary. The turning angle can be random. After turning, it continues to move until it encounters a boundary again, repeating the steps of turning at a random angle. Furthermore, the turning angle can be randomized within a certain range to avoid being unable to move away from the current boundary due to an angle that is too small or too large.

[0200] Furthermore, based on the map and the machine's real-time positioning, a specific turning angle is determined so that the machine can turn and move from its current location on the target boundary segment to another target boundary segment.

[0201] In some embodiments, controlling the autonomous working machine 10 to move and work within a working area according to a first map includes: controlling the autonomous working machine 10 to move and work within a work area based on a semi-random movement mode, wherein the work area is a part or all of the working area.

[0202] The autonomous working machine 10 can move semi-randomly throughout the entire work area or within a department area.

[0203] The autonomous working machine 10 moves within the working area based on a semi-random movement pattern and performs feature acquisition. For example, the specific process can be found in Embodiments 7 to 9 below. Furthermore, it can be combined with a method of moving along the boundary and performing feature acquisition.

[0204] In some embodiments, controlling the autonomous working machine 10 to move and work within a working area according to a first map includes: determining multiple reversing positions on the boundary of the working area according to the first map;

[0205] The autonomous working machine 10 is controlled to move and work within the working area in a semi-random movement mode. In the semi-random movement mode, the autonomous working machine 10 moves to a reversing position, turns at the reversing position, and moves to other reversing positions.

[0206] Based on the boundary coordinates in the first map, multiple reversal positions are determined, and these positions are evenly distributed along the boundary of the work area. When moving in a semi-random movement mode, the autonomous working machine 10 starts from a certain position and moves towards the boundary, specifically towards any one of the multiple reversal positions. Then, the autonomous working machine 10 turns at the reversal position to avoid going out of bounds, and then moves towards other reversal positions.

[0207] Each reversal position is located on or near the boundary, and the autonomous working machine 10 can perform feature supplementation at the reversal position.

[0208] Furthermore, the autonomous working machine 10 performs feature acquisition by means of the turning process at the reversing position.

[0209] Furthermore, when turning at a reversing position, the turning direction can be uniformly set to either right or left; the turning angle can be randomly determined within a preset angle range to ensure that more first-update images from different perspectives can be acquired during the turning process; to this end, the turning angle can be randomly determined within a preset angle range based on the machine's current position, current direction of movement, and relative positional relationship with other reversing positions.

[0210] The autonomous working machine 10 moves within the working area based on a semi-random movement pattern, moving between multiple reversing positions and performing feature supplementation. For example, the specific process can be found in Embodiment 10 below.

[0211] In some embodiments, the autonomous working machine 10 moves within the working area based on a fully random movement mode and performs feature acquisition. In the fully random or semi-random movement mode, the machine automatically turns and moves to other boundary positions when it encounters a boundary; the difference between the two modes is that in the fully random movement mode, the turning angle is not determined based on the relative positional relationship between the current position, the current direction of movement and other reversing positions, but is determined completely randomly within a certain angle range.

[0212] After obtaining the first map, in step 305, the autonomous working machine 10 moves according to the first map and only performs feature supplementation at the target location, which can improve the feature supplementation efficiency.

[0213] In some embodiments, step 305, the method for determining that the autonomous working machine 10 has moved to the target position includes: determining the machine position of the autonomous working machine 10 and the first information corresponding to the machine position during the movement, the first information including at least one of the following parameters: visual positioning accuracy, distance between the machine position and the previous target position, and time interval between the current time point and the time point when the machine moves to the previous target position; and determining that the autonomous working machine 10 has moved to the target position in response to the first information satisfying a first condition.

[0214] When the autonomous working machine 10 moves according to the first map, it uses VSLAM technology to determine the specific coordinates of its position in real time. However, the positioning accuracy of VSLAM is affected by the first map and the real-time images acquired by the machine. If a visual loop closure can be successfully detected at a certain location, it indicates that the visual positioning accuracy at that location is high and meets the first condition, so that location can be used as the target location.

[0215] If the machine's location is very close to the previous target location and an image is acquired at that location, the image will be very similar to the first updated image acquired at the target location due to the proximity. The resulting visual features will also be similar. Therefore, it's unnecessary to perform feature supplementation at a location very close to the target location; otherwise, massive amounts of redundant data would be generated, consuming processing and memory resources and reducing processing speed. Therefore, it's necessary to determine the distance between the machine's location and the previous target location. If the distance is greater than a set distance value, the first condition is met, and that location can be used as the target location. The set distance value can be adjusted according to the actual situation.

[0216] Similarly, if the current time point is close to the time point corresponding to the previous target location, the image captured at the current time point will be very similar to the first updated image captured at the previous target location, and the resulting visual features will also be similar. Therefore, it is unnecessary to perform feature supplementation at a time point very close to the time point corresponding to the target location; otherwise, massive amounts of redundant data will be generated, consuming processing and memory resources and reducing processing speed. Therefore, it is necessary to determine the time interval between the current time point and the time point at which the target location was last determined. If the time interval is greater than a set duration value, the first condition is met, and the machine location at the current time point can be used as the target location. The set duration value can be adjusted according to the actual situation.

[0217] In some embodiments, determining that the autonomous working machine 10 has moved to the target location in response to the first information satisfying the first condition includes: determining that the autonomous working machine 10 has moved to the target location in response to the first information satisfying the first condition and the machine location being located in an unclear boundary segment.

[0218] Since machines are more prone to problems such as going out of bounds in unclear boundary sections, it is of great importance to supplement the feature sampling of unclear boundary sections to improve the accuracy of subsequent positioning.

[0219] Furthermore, to save time spent on feature acquisition, feature acquisition can be performed only in unclear boundary sections. For example, when the autonomous machine 10 moves along the boundary to acquire features, feature acquisition can be performed only at locations in unclear boundary sections where the first information meets the first condition, thereby saving time spent moving along the boundary and allowing the machine to enter the area for work as soon as possible.

[0220] In some embodiments, determining that the autonomous working machine 10 has moved to the target location in response to the first information satisfying the first condition includes: determining the machine location as the target location in response to the machine location being on or near a boundary and the positioning accuracy of the machine location meeting the visual loop closure condition.

[0221] The first condition includes the machine being located on or near the boundary and a visual loop being detected.

[0222] Alternatively, the first condition includes the machine position being on or near the boundary, a visual loop being detected, and the distance between the machine position and the previous target position being greater than a distance set value.

[0223] After determining the current location as the target location, the first updated image is taken at the current location. The first updated image needs to reflect as many features of the surrounding environment as possible. The following method can be used to control the autonomous working machine 10 to make the features corresponding to the first updated image richer.

[0224] In some embodiments, acquiring a first updated image at a target location includes: controlling the autonomous working machine 10 to rotate at the target location, and during the rotation, acquiring a first updated image based on the camera of the autonomous working machine 10, wherein the first updated image includes at least two images from different angles.

[0225] After determining the current location as the target location, the relative positional relationship between the camera and the machine body remains unchanged. The autonomous working machine 10 is controlled to rotate or circle at the current location, and the camera acquires a first updated image during the machine's rotation. The number of first updated images can be set according to actual conditions; the more images, the more visual features can be obtained. Different shooting angles for the first updated images are beneficial for capturing different environmental objects and obtaining rich visual features. For example, the specific process can be found in the relevant description in Embodiment 1 below. Other embodiments can also adopt this method when performing feature acquisition.

[0226] In some embodiments, acquiring a first updated image at a target location includes: controlling a camera to rotate relative to the body of the autonomous working machine 10, and acquiring a first updated image based on the camera of the autonomous working machine 10 during the camera rotation, wherein the first updated image includes at least two images from different angles.

[0227] After determining the current position as the target position, the autonomous working machine 10 is kept in the same pose, while the camera is rotated relative to the body of the autonomous working machine 10, and the first updated image is captured during the camera rotation.

[0228] In some embodiments, acquiring a first updated image at a target location includes: controlling the panoramic camera of the autonomous working machine 10 to acquire a panoramic image, wherein the first updated image is a panoramic image.

[0229] If a panoramic camera is used, a large area of ​​the environment can be captured at once.

[0230] The three methods of acquiring the first updated image described above are not only applicable to feature acquisition in scenarios where the machine moves and / or works along the boundary, but also applicable to feature acquisition in scenarios where the machine moves and works within the working area.

[0231] In addition to the three acquisition methods mentioned above, other acquisition methods can also be used when the machine moves and works within the work area. For example, in some embodiments, acquiring the first updated image at the target location includes: acquiring the first updated image through the camera of the autonomous working machine 10 at the target location where the autonomous working machine 10 turns based on the first map.

[0232] When the autonomous machine 10 is within the work area, it moves and works according to the first map and based on a planned or semi-random movement mode. If the machine moves near the boundary of the work area, in order to stay within the boundary and ensure safety, it needs to be controlled to turn away from the boundary. During the turning process, the camera's viewing angle changes with the machine's turn, allowing the machine to capture the first updated image. This efficient acquisition of the first updated image is achieved without the user's awareness, demonstrating the machine's intelligence and smooth operation. For example, the specific process can be found in Embodiments 1, 4, 6, 7, 8, 9, and 10 below.

[0233] Specifically, at the target location where the autonomous machine 10 turns based on the first map, the first updated image is acquired through the camera of the autonomous machine 10, including:

[0234] When the autonomous working machine 10 turns in place along the planned path based on the first map, it acquires the first updated image through its camera. While the autonomous working machine moves or works along the planned path, it starts from the boundary position and moves towards the interior of the working area, cutting through it. After reaching the opposite boundary (or near the boundary, for example, within 0.2 meters of the boundary), it turns and moves a distance (e.g., 0.4 meters) along the boundary. During this turning process, the camera acquires the first updated image. After moving a further distance along the boundary, the autonomous working machine continues to turn towards the interior of the working area, and during this turning process, the camera acquires the first updated image.

[0235] The aforementioned in-situ turning includes: controlling the autonomous working machine 10 to stop moving forward; and then controlling the autonomous working machine 10 to rotate around a preset rotation center by a preset angle, wherein the preset angle is less than or equal to 180 degrees. The aforementioned in-situ turning operation is a precise pose adjustment strategy, and its execution process includes the following steps:

[0236] First, the control circuit of the autonomous working machine sends a command to the drive unit to stop the current forward or moving action of the autonomous working machine 10, bringing it to a standstill. This is intended to provide a stable starting point for subsequent orientation adjustments, avoiding positioning errors or path deviations that may occur when turning during movement.

[0237] Then, the control circuit determines a preset rotation center and a preset angle based on the built-in navigation algorithm or real-time path planning requirements. Subsequently, by precisely controlling the differential speed of the left and right wheel sets in the drive unit (i.e., the left wheel set and the right wheel set rotate at speeds of equal magnitude but opposite directions), the autonomous working machine is driven to rotate in place around the preset rotation center (O).

[0238] The preset rotation center is a virtual point set according to the machine structure and working logic, and its position can be dynamically defined according to different task scenarios. For example, in a preferred embodiment, the rotation center is located on the geometric centerline of the robot; in another embodiment, in order to achieve a U-turn with a smaller turning radius, the rotation center can be set on the drive wheel on one side of the robot, thereby achieving "zero radius" turning with a single wheel as the axle.

[0239] The preset angle is typically less than or equal to 180 degrees. This angle range is set based on optimization considerations of work efficiency and behavioral logic: small-angle turns (such as 15° to 45°) are often used for fine-tuning during movement to align with boundaries; while 90° or 180° turns are often used to achieve direction switching or complete U-turns. For example, when the machine detects a non-lawn area ahead, it can trigger a turn of approximately 180 degrees, causing its operation to face the opposite direction, thus continuing to mow the lawn in work area A. Limiting the maximum turning angle to within 180 degrees avoids unnecessary 360-degree full rotation, effectively improving work efficiency and reducing energy consumption.

[0240] Ultimately, through this series of orderly control commands, the autonomous working machine 10 is able to efficiently and accurately adjust its orientation without the need for large displacements, while simultaneously capturing the first updated image through the camera.

[0241] After acquiring the first updated image, the autonomous machine 10 updates the first map using the first updated image. In some embodiments, the autonomous machine 10 can obtain the inertial navigation information corresponding to the first updated image and determine the visual features of the first updated image. The aforementioned inertial navigation information may include information such as position, yaw angle, and velocity, mainly reflecting the pose of the autonomous machine 10. The visual features extracted from the image may include, but are not limited to, one or any combination of: feature points, descriptive information of feature points, position information of feature points in the image, three-dimensional coordinate information of feature points, edges, textures, histograms, etc.

[0242] Then, the descriptive information of the feature points, their position information in the image, and their 3D coordinate information are determined. Based on VSLAM technology, the first map is updated according to the inertial navigation information, the descriptive information of the feature points, their position information in the image, and their 3D coordinate information. The first map can be updated by directly modifying, adding, or deleting information on the map, or by generating a new map based on the existing map.

[0243] In some embodiments, the autonomous working machine 10 is controlled to operate before it moves according to the first map and performs feature acquisition; or, the autonomous working machine 10 is controlled to operate while it moves according to the first map and performs feature acquisition; or, the autonomous working machine 10 is controlled to operate after it moves according to the first map and performs feature acquisition; wherein, controlling the autonomous working machine 10 to move according to the first map includes controlling the autonomous working machine 10 to move along a boundary according to the first map, or controlling the autonomous working machine 10 to move within a working area according to the first map.

[0244] There are several possible operating sequences for the autonomous working machine 10. It can work first and then perform feature acquisition, for example, mapping → autonomous working machine 10 works → feature acquisition along the boundary → autonomous working machine 10 works; it can also perform feature acquisition while working, for example, mapping → feature acquisition while working → autonomous working machine 10 works, mapping → feature acquisition along the boundary → feature acquisition while working; or it can perform feature acquisition first and then work, for example, mapping → feature acquisition along the boundary → work.

[0245] In some embodiments of this application, after acquiring the first map, the scheme of controlling the autonomous working machine 10 to perform feature acquisition on the first map during movement can employ various runtime sequences. For example: "Mapping → Feature acquisition along the boundary → Working", "Mapping → Feature acquisition while working → Autonomous working machine 10 working", "Mapping → Feature acquisition along the boundary → Feature acquisition while working", "Mapping → Autonomous working machine 10 working → Feature acquisition along the boundary → Autonomous working machine 10 working", etc. Among these various runtime sequences, one scenario involves controlling the autonomous working machine 10 to move according to the first map, performing feature acquisition during the movement. Another scenario involves controlling the autonomous working machine 10 to move and work according to the first map, performing feature acquisition while moving and working.

[0246] The following detailed examples of various runtime sequences are provided using different implementation schemes.

[0247] In some embodiments, the autonomous working machine 10 performs tasks and feature acquisition sequentially, interleavedly, or simultaneously based on a first map. For example, the autonomous working machine 10 performs tasks after feature acquisition, or performs feature acquisition after performing tasks, or performs feature acquisition during the process of performing tasks.

[0248] In some embodiments, the autonomous working machine 10 first performs work (i.e., executes partial work tasks, such as performing work in a partial sub-region of the work area), and during and / or after the work, performs feature supplementation (the specific process can be found in embodiments five, six, seven, eight, and nine below). The aforementioned partial work tasks can be work tasks within a workable area formed between several boundary segments. Exemplarily, the aforementioned boundary segments are clear boundaries.

[0249] In some embodiments, the autonomous working machine 10 may first perform feature supplementation (which may be feature supplementation of partial boundaries) before starting work (the specific process can be found in Embodiments 2, 3, 4, and 10 below). Exemplarily, the initial feature supplementation may be for clear boundary segments, unclear boundary segments, or boundary segments without limitations, and feature supplementation can be performed through a path scheme that is entirely along the edge or partially along the edge. Exemplarily, partial sub-regions formed between the boundary segments where feature supplementation has been performed serve as workable regions, where work can be performed (the specific process can be found in Embodiments 2, 3, and 5 below). In these workable regions, the autonomous working machine 10 employs a planned mode or a semi-random mode, such as an automatic cutter performing planned cutting or semi-random cutting. If feature supplementation of a partial boundary is not completed, the autonomous working machine 10 can continue to perform feature supplementation tasks for that partial segment after work has commenced. However, it is feasible for the autonomous working machine 10 to perform feature supplementation during the work process; for example, when the autonomous working machine 10 moves to the boundary, it determines a target location on the boundary and performs feature supplementation. For example, after partial boundary feature acquisition, a globally usable area can be formed if certain conditions are met (Examples 4 and 10). In this case, the autonomous working machine 10 performs work within the working area based on a planned path or a random path, and performs feature acquisition tasks during and / or after the work. The aforementioned specific conditions may be: feature acquisition of unclear segments has been completed.

[0250] Furthermore, in some embodiments, the target location for performing the feature acquisition task meets predetermined conditions. For example, the target location can form a visual loop.

[0251] Solution Example 1

[0252] This scheme adopts the following runtime sequence: "Map building → Feature acquisition along the boundary → Autonomous working machine 10 working".

[0253] After obtaining the first map, the autonomous working machine 10 can be controlled to move along the boundary according to the first map. During the movement of the autonomous working machine 10 along the boundary, feature supplementation is performed until it returns to the starting position of the movement, or the distance between the machine position of the autonomous working machine 10 and the starting position is less than the distance threshold, and the feature supplementation ends.

[0254] The aforementioned feature acquisition refers to acquiring images (referred to as "first updated images" in this embodiment) at each target location on the boundary, and using the first updated images to update the first map. The first updated images represent images corresponding to the surrounding environment of the boundary. The aforementioned target locations can be located on the boundary or within a preset distance from the boundary.

[0255] As one possible approach, adjacent target locations can be spaced at a preset distance, for example, feature sampling can be performed every 2 meters.

[0256] As another feasible approach, a preset time interval can be used between the acquisition times of adjacent target locations, for example, feature acquisition can be performed every 10 seconds.

[0257] As another feasible approach, unclear boundary locations can also be used as target locations for feature acquisition. It's important to note that after mapping is completed, semantic information for the first map can be generated, describing the clarity of the boundaries. Clarity can mean: when the autonomous machine 10 is located at an unclear boundary location, it cannot identify the boundary location based solely on the image; when the autonomous machine 10 is located at a clear boundary location, it can identify the boundary location based solely on the image. In other words, if the autonomous machine 10 cannot identify the boundary location using image recognition technology (such as semantic segmentation), it indicates that the boundary at that location is unclear, and the machine is prone to going out of bounds. Feature acquisition is then necessary—that is, acquiring a first updated image at that boundary location and extracting visual features to update the first map, increasing the visual features near the unclear boundary location, thereby improving the machine's positioning accuracy near unclear boundary locations and preventing it from going out of bounds.

[0258] In other words, the autonomous machine 10 moves along the boundary and acquires the first updated image at points on the boundary or at points within a preset distance from the boundary. It obtains the inertial navigation information corresponding to the first updated image and determines the feature points of the first updated image. Then, it determines the descriptive information of the feature points, their position information in the image, and their three-dimensional coordinate information. Based on VSLAM technology, it updates the first map according to the inertial navigation information, the descriptive information of the feature points, their position information in the image, and their three-dimensional coordinate information. This process is similar to the mapping process and will not be elaborated here. Updating the first map can be done by directly modifying, adding, or deleting information on the map, or by generating a new map based on the existing map.

[0259] When acquiring the first updated image during the feature acquisition process, the following methods can be used, but are not limited to:

[0260] The first method involves controlling the autonomous working machine 10 to rotate at the target position, and during the rotation, acquiring a first updated image based on the camera of the autonomous working machine 10. The first updated image may include at least two images from different angles.

[0261] For example, as an automatic lawnmower moves along a boundary, it rotates in place at its current position every certain distance, while the camera remains stationary, thus enabling image acquisition from multiple directions at that location.

[0262] The second method involves controlling the camera to rotate relative to the body of the autonomous working machine 10, and during the rotation of the camera, acquiring a first updated image based on the camera of the autonomous working machine 10. The first updated image includes at least two images from different angles.

[0263] For example, every time an automatic lawnmower moves a certain distance, it controls the camera to rotate while the lawnmower remains stationary, thus enabling image acquisition from multiple directions.

[0264] The third method: control the panoramic camera of the autonomous working machine 10 to collect panoramic images, and the first updated image is the panoramic image.

[0265] For example, by equipping an automatic lawnmower with a panoramic camera, the camera can capture panoramic images every time the lawnmower moves a certain distance, thus enabling image acquisition from multiple directions.

[0266] Besides the three methods mentioned above, other methods can also be used, which will not be listed here. Furthermore, in subsequent embodiments, the methods for acquiring the first updated image during feature acquisition can all employ the three methods described above, which will not be elaborated upon in later embodiments.

[0267] Additionally, when ending feature acquisition, one possible approach is to record the starting position of the movement, i.e., the starting position of feature acquisition. If the system repositions to that starting position or to a position less than a preset distance threshold, it indicates that the autonomous machine 10 has completed one cycle around the boundary of the working area, and the acquisition can be terminated. The starting position can be determined in any way, such as positioning based on a positioning module, positioning based on VSLAM, etc.

[0268] As another possible approach, an auxiliary device such as a limit switch can be set at the starting position of the movement. If the autonomous machine 10 returns to the limit switch, it means that the autonomous machine 10 has completed one cycle around the boundary of the working area, and the re-sampling can be completed. Other implementation methods can also be used, which will not be listed here.

[0269] like Figure 4a As shown, after completing the mapping process, the automatic lawnmower obtains the first map of the boundary of the work area A. Starting from the boundary position a of the work area A, the automatic lawnmower moves along the boundary, performing feature re-sampling every 2 meters during the movement, until it returns to the boundary position a. Figure 4a The orange arrows on the boundary represent the movement path corresponding to feature acquisition, and the circles on the boundary represent the location of feature acquisition, i.e., the target location.

[0270] Furthermore, during the aforementioned movement and feature acquisition along the boundary, boundary-related tasks can also be performed simultaneously. For example, an automated lawnmower can perform feature acquisition and cutting simultaneously while moving along the boundary. This approach improves the efficiency of the autonomous working machine 10. Moreover, allowing the autonomous working machine 10 to begin work earlier effectively shortens the waiting time between the creation of the first map and the start of the autonomous moving machine's work, thus enhancing the user experience.

[0271] After completing the aforementioned feature acquisition, the autonomous working machine 10 can utilize the updated first map to operate within the work area. For example, an automatic lawnmower can mow the grass within the work area. Figure 4b As shown, a "bow-shaped cutting" method can be used to cut within work area A, with the cutting path indicated by black arrows. For example, after moving 0.4 meters along the boundary, the path turns inward towards the work area and cuts through it. Upon reaching the opposite boundary, the path turns again, moves 0.4 meters along the boundary, and then turns inward again towards the work area and cuts through it, continuing this process to complete the mowing of the work area. During the process, the coordinates of the boundary positions in the first map and visual features are used to determine whether the boundary has been reached and how to move along it. No additional feature acquisition is performed during the process.

[0272] Scheme Example 2

[0273] This scheme adopts the following runtime sequence: "Map building → Feature acquisition along the boundary → Autonomous working machine 10 working".

[0274] After acquiring the first map, the autonomous machine 10 can be controlled to move along the boundary based on the first map, and feature acquisition can be performed during the movement of the autonomous machine 10 along the boundary. The boundary segment traversed by the feature acquisition is determined; in response to the boundary segment forming a first sub-region, the autonomous machine 10 is controlled to work in the first sub-region; in response to the completion of the work of the autonomous machine 10 in the first sub-region, the process returns to the first map, controlling the autonomous machine 10 to move along the boundary until the autonomous machine 10 returns to the starting position of the movement, which is the starting position of the feature acquisition.

[0275] like Figure 5a As shown, taking an automatic lawnmower as an example, after completing the mapping process, the automatic lawnmower obtains the first map of the boundary of the work area A. Starting from the boundary position b5 of the work area A, the automatic lawnmower moves along the boundary, performing feature acquisition every 2 meters during the movement (the orange arrows in the figure represent the feature acquisition path, and the circles represent the feature acquisition positions, i.e., the target positions). If the boundary segment traversed by the feature acquisition forms a sub-region (referred to as the first sub-region in this embodiment), for example, recording the leftmost boundary position P1 and the right boundary position P2 of the currently "enclosed" semi-enclosed sub-region, if the distance d between P1 and P2 (this distance can be a distance in a preset direction, represented by red lines and red font d in the figure) reaches a preset threshold m, then feature acquisition stops. The preset threshold m can be several meters or tens of meters. A "bow-shaped cut" is performed within the currently "enclosed" semi-enclosed sub-region (the cutting path is represented by black arrows). After the cutting is completed, as shown... Figure 5b As shown in the diagram, the system continues to move along the boundary from position P2 and perform feature acquisition (the orange arrows in the diagram represent the feature acquisition path). Once a new first sub-region is formed, for example, the distance s between P3 and P2 (represented by yellow lines and yellow text s in the diagram) reaches a preset threshold m, feature acquisition stops. A "bow-shaped cut" is then performed within the currently enclosed first sub-region (the cutting path is represented by black arrows). After the cut is completed, the system continues to move along the boundary from position P3 and perform feature acquisition, and so on, until it returns to the initial position b5, completing feature acquisition for all target positions on the boundary. This forms a runtime sequence of "mapping → feature acquisition along the boundary → autonomous machine 10 working → feature acquisition along the boundary → autonomous machine 10 working…".

[0276] The processing performed during feature acquisition is similar to that in Embodiment 1 of the above scheme, that is, an image (referred to as the "first updated image" in this embodiment) is acquired at the target location, and the first map is updated using the first updated image. The first updated image represents the image corresponding to the surrounding environment of the boundary. The target location can be located on the boundary or within a preset distance from the boundary. As for the selection of the target location and the specific method of acquiring the first updated image, please refer to the relevant description in Embodiment 1 of the scheme, which will not be repeated here.

[0277] This approach first enhances the robustness of the autonomous mobile machine's localization near the boundary by supplementing features along the boundary, resulting in richer visual features that make it less likely for the autonomous mobile machine to lose its localization when approaching the boundary. Secondly, by combining feature supplementation along the boundary with starting operation after the first sub-region is formed, the waiting time for the user to access the autonomous mobile machine is reduced. This shortens the waiting time from the creation of the first map to the operation of the autonomous mobile machine, thus improving the user experience.

[0278] Scheme Example 3

[0279] This scheme adopts the following runtime sequence: "Map building → Feature acquisition along the boundary → Autonomous working machine 10 working".

[0280] After obtaining the first map, the autonomous working machine 10 can be controlled to move along the boundary based on the first map, and feature supplementation can be performed during the movement of the autonomous working machine 10 along the boundary.

[0281] The acquired first map may include boundary information corresponding to multiple boundary segments. When controlling the autonomous machine 10 to move along the boundaries, a second sub-region can be determined based on the boundary information corresponding to the multiple boundary segments included in the first map; the autonomous machine 10 is then controlled to move along the boundary segments of the second sub-region, and feature acquisition is performed during the movement. In response to completing the feature acquisition of the boundary segments of the second sub-region, the autonomous machine 10 is controlled to move and operate within the second sub-region.

[0282] In this embodiment, the aforementioned boundary segment can be a clear boundary segment. When the autonomous machine 10 is located at a clear boundary position, it can identify the boundary position solely based on the image. That is, if the autonomous machine 10 can identify the boundary position using visual features in the image, it indicates that the image at that boundary position is clear. The purpose of feature supplementation at clear boundaries is to enrich the visual features of the clear boundaries, thereby improving the positioning accuracy at the boundaries. Feature supplementation involves acquiring a first updated image at the target position and extracting visual features from it to update the first map.

[0283] like Figure 6a As shown in the diagram, taking an automatic lawnmower as an example, after completing the mapping process, the automatic lawnmower obtains a first map of the boundary of the work area A. Clear boundary segments are found on the boundary of the work area in the first map. The automatic lawnmower is less likely to go out of bounds on these clear boundary segments. Therefore, if multiple clear boundary segments can form a sub-region (referred to as the second sub-region in this embodiment), for example… Figure 6a As shown, the two clearly defined boundary segments a6-b6 and c6-d6 can form a semi-closed region. We can first move along the clearly defined boundary segments a6-b6 and c6-d6, performing feature re-sampling every 2 meters during the movement (the orange arrows in the diagram represent the feature re-sampling path, and the circles represent the feature re-sampling positions, i.e., the target positions). After re-sampling is completed, a "bow-shaped cut" is performed in the second sub-region formed by segments a6-b6 and c6-d6 (the cutting path is indicated by black arrows). During the cutting process, no feature re-sampling is performed; only the cutting operation is performed.

[0284] Furthermore, the aforementioned boundary segments can also be unclear boundary segments. Specifically, when the autonomous machine 10 is located at an unclear boundary position, it cannot identify the boundary position based solely on the image. In other words, if the autonomous machine 10 cannot identify the boundary position using only the image at a certain boundary location, it indicates that the image at that boundary position is unclear. The purpose of feature acquisition at unclear boundaries is to improve the relocalization success rate of unclear boundaries. Feature acquisition involves acquiring a first updated image at each target location within the unclear boundary and extracting visual features from it to update the first map.

[0285] like Figure 6b As shown in the diagram, an automatic lawnmower is used as an example. After performing a "bow-shaped cut" in the second sub-region formed by segments a6-b6 and c6-d6, it begins to move along the unclear boundary segment db, and features are supplemented during the movement (the orange arrows in the diagram represent the feature supplementation path, and the circles represent the feature supplementation positions, i.e., the target positions). After the supplementation is completed, a "bow-shaped cut" is performed in the second sub-region formed by segments d6-b6 (segments d6-b6 form a semi-closed sub-region) (the cutting path is indicated by green arrows). During the cutting process, no feature supplementation is performed; only the cutting operation is carried out.

[0286] After the second sub-region is cut, it can continue to move along the unclear boundary segment a6-c6. During the movement, feature supplementation is performed, and after the supplementation is completed, "bow-shaped cutting" is performed in the second sub-region formed by segment a6-c6. This process is similar to the previous process and will not be described in detail.

[0287] The processing performed during feature acquisition is similar to that in Embodiment 1 of the above scheme, that is, an image (referred to as the "first updated image" in this embodiment) is acquired at the target location, and the first map is updated using the first updated image. The first updated image represents the image corresponding to the surrounding environment of the boundary. The target location can be located on the boundary or within a preset distance from the boundary. As for the selection of the target location and the specific method of acquiring the first updated image, please refer to the relevant description in Embodiment 1 of the scheme, which will not be repeated here.

[0288] It can be seen that the above process is equivalent to "Mapping → Feature acquisition along the boundary → Autonomous working machine 10 working → Feature acquisition along the boundary → Autonomous working machine 10 working...".

[0289] This approach first enhances the robustness of the autonomous mobile machine's localization near the boundary by supplementing features along the boundary, resulting in richer visual features that make it less likely for the autonomous mobile machine to lose its location when approaching the boundary. Secondly, it begins operation after forming the second sub-region through feature supplementation along the boundary, reducing the user's waiting time for the autonomous mobile machine to start working. This shortens the waiting time from the creation of the first map to the start of autonomous mobile machine operation, thus improving the user experience.

[0290] Scheme Example 4

[0291] This solution adopts a runtime sequence of "map creation → feature acquisition along the boundary → feature acquisition while working".

[0292] After obtaining the first map, the autonomous working machine 10 can be controlled to move along the boundary based on the first map, and feature supplementation can be performed during the movement of the autonomous working machine 10 along the boundary.

[0293] The first map mentioned above includes boundary information corresponding to unclear boundary segments. Accordingly, the autonomous machine 10 can first be controlled to move along the unclear boundary segments, performing feature acquisition during the movement until all unclear boundary segments have been acquired. Then, the autonomous machine 10 is controlled to work throughout the entire working area, turning when encountering boundaries and simultaneously acquiring features at those boundaries. Since the boundaries of all working areas can be considered clear after all unclear boundary segments have been acquired, the feature acquisition performed when turning at boundaries while controlling the autonomous machine 10 to work throughout the entire working area is actually for the clear boundary locations. The purpose is to enrich the visual features at the boundaries, making it less likely for the autonomous machine to lose its position when approaching the boundaries, thereby improving the robustness of the autonomous machine's positioning near the boundaries.

[0294] like Figure 7aAs shown, after mapping is completed, the first map of the boundary of working area A is obtained. Segments a7-f7, b7-c7, and d7-e7 are clear boundary segments, while segments a7-b7, c7-d7, and e7-f7 are unclear boundary segments. First, the autonomous working machine 10 is controlled to move along the boundary, performing feature acquisition in the unclear boundary segments a7-b7, c7-d7, and e7-f7. The orange arrows in the figure represent the feature acquisition path, and the circles represent the feature acquisition locations, i.e., the target locations.

[0295] The process of controlling the autonomous machine 10 to move along an unclear boundary segment may include: starting from a boundary position where the positioning accuracy meets the visual loop closure condition, and according to a first map, controlling the autonomous machine 10 to move along the unclear boundary segment. The aforementioned visual loop closure condition refers to the autonomous machine 10 recognizing that it has previously reached that boundary position. To determine whether a boundary position meets the visual loop closure condition, an image can be acquired at that boundary position, and the acquired image can be matched with the image at that position in the first map (e.g., the similarity is greater than or equal to a preset similarity threshold). If the match is successful, the visual loop closure condition is considered met. The matching process primarily utilizes visual features extracted from the images.

[0296] Boundary locations that typically satisfy the visual closure condition are sharp boundary locations. For example... Figure 7a As shown, movement can begin from a point on a clearly defined boundary segment, such as a point on the boundary between a7 and f7. Movement can also begin from the boundary between a clearly defined and an unclear boundary segment, such as point a7.

[0297] In this scheme, after completing feature acquisition in unclear boundary sections, the autonomous working machine 10 continues to acquire features while working throughout the entire working area based on a planned movement mode.

[0298] In one embodiment, the autonomous working machine performs feature acquisition on the first map in at least some rounds of work tasks, and the working path of the autonomous working machine (10) is different in work tasks in which feature acquisition is performed and in work tasks in which feature acquisition is not performed.

[0299] In other words, if feature acquisition is performed simultaneously through multiple rounds of tasks, different work paths can be used in different rounds of tasks. These different rounds of tasks include both overall tasks and sub-regional tasks. A global task refers to controlling the autonomous machine 10 to move and work within a work area based on a planned movement mode. A sub-regional task refers to controlling the main machine 10 to move and work within sub-regions divided into work area A based on a planned movement mode. The planned movement mode refers to the mode in which the autonomous machine 10 uses a path planning module to plan its path for movement and work.

[0300] like Figure 7b As shown, after completing feature acquisition for unclear boundary sections, the automatic lawnmower is controlled in work area A using a planning mode to perform a "bow-shaped" cut (the cutting path is indicated by black arrows). During the cutting process, the lawnmower performs cutting operations within work area A, for example, cutting from the boundary position inwards and through the work area. Upon reaching the opposite boundary (or near the boundary, for example, within 0.2 meters), it turns and moves a distance (e.g., 0.4 meters) along the boundary before turning again inwards and through the work area, and so on, thus completing the mowing work in the work area. During the operation, the coordinate information of the boundary position in the first map and / or image recognition results are used to determine whether the boundary has been reached and how to move along it. When reaching the boundary, the lawnmower moves a distance along the boundary, and then feature acquisition is performed when turning, indicated by purple arrows in the figure. From... Figure 7b As can be seen, since the machine builds the map along the edge in a counter-clockwise direction, when building the map at the upper boundary, the machine moves from the right side of the working area to the left side, and its perspective is also from right to left; when building the map at the lower boundary, the machine moves from the left side of the working area to the right side, and its perspective is also from left to right; when performing the first global task, the working direction is from the left side of the working area to the right side, so at the upper boundary, the machine moves from the left side of the working area to the right side, and this perspective is opposite to the perspective during map building, making it difficult to generate visual loops, and the machine has difficulty performing feature supplementation at the upper boundary, while at the lower boundary, the machine's perspective is the same as during map building, and the machine can perform feature supplementation at the lower boundary; furthermore, a second global task needs to be performed (such as Figure 7c The working direction is from the right side of the working area to the left side, allowing the machine to perform feature acquisition at the upper boundary. Therefore, considering the influence of direction, it can be done according to... Figure 7b After performing a round of global cutting in the "bow-shaped" direction as shown, perform another round of global cutting within the work area A in the opposite direction, as follows. Figure 7cAs shown in the figure, feature acquisition is also performed when turning at the boundary during the cutting process, thereby ensuring the accuracy of feature acquisition.

[0301] Since feature acquisition has already been performed on unclear boundary sections, feature acquisition can be performed only when turning at clear boundaries during the operation. Alternatively, feature acquisition can be performed at any boundary when turning.

[0302] This approach first enhances the robustness of the autonomous mobile machine's localization near the boundary by supplementing features along unclear boundary sections, making it less prone to losing its location when approaching the boundary. Furthermore, after completing feature supplementation for unclear boundary sections, the entire working area is processed, reducing the user's waiting time for the autonomous mobile machine to begin operation. This shortens the waiting time from the initial map creation to the autonomous mobile machine's operation, improving the user experience. Moreover, further feature supplementation is performed at the boundary during the process, resulting in richer visual features at the boundary locations.

[0303] Similar to the global task described above, the sub-region task involves dividing the entire work area into multiple sub-regions. Within each sub-region, an automated lawnmower is controlled to perform a "bow-shaped" cut based on a planning mode. The movement and feature acquisition methods of the autonomous working machine are similar to those of the global task, and will not be repeated here.

[0304] Regarding the above embodiments, it should also be noted that the difference in the working path of the autonomous machine 10 between a task involving feature acquisition and a task without feature acquisition includes at least one of the following:

[0305] The path spacing of the work path differs between work tasks that perform feature acquisition and those that do not. For example, ... Figure 7b As shown, in a certain round of work, the autonomous machine does not perform feature acquisition. Starting from the boundary, the machine cuts towards the interior of the work area, passing through it. Upon reaching the opposite boundary (or near the boundary, for example, within 0.2 meters), it turns and moves a short distance along the boundary (e.g., 0.4 meters), then continues cutting towards the interior of the work area, repeating this process to complete the mowing. In the next round of work, the autonomous machine performs feature acquisition. The distance it moves along the boundary differs from the previous round; it can be set larger (e.g., 0.6 meters, 0.8 meters) or smaller (e.g., 0.2 meters, 0.3 meters).

[0306] The path direction of the workflow differs between tasks that perform feature acquisition supplementation and those that do not. For example... Figure 7b As shown, in one round of tasks, the autonomous working machine does not perform feature acquisition, and its working direction is from the left side to the right side of the working area. In the next round of tasks, the autonomous working machine performs feature acquisition, and its working direction is from the right side to the left side of the working area. Alternatively, in one round of tasks, the autonomous working machine does not perform feature acquisition, and its working direction is from the right side to the left side of the working area. In the next round of tasks, the autonomous working machine performs feature acquisition, and its working direction is from the left side to the right side of the working area.

[0307] The autonomous machine 10 travels at different speeds along the work path during and without feature acquisition tasks. For example, in a certain round of work, if the autonomous machine does not acquire additional features, its movement speed is 2 m / s. In the next round of work, if the autonomous machine acquires additional features, its movement speed can be set to 1 m / s. The autonomous machine's movement speed can be faster or slower during feature acquisition tasks than during tasks without feature acquisition.

[0308] Scheme Example 5

[0309] This scheme adopts the following runtime sequence: "Map building → Autonomous worker 10 working → Feature acquisition along the boundary → Autonomous worker 10 working".

[0310] After obtaining the first map, the autonomous working machine 10 can be controlled to move and work within the working area based on the planned movement mode, according to the first map. The working area is a part of the working area.

[0311] The first map mentioned above includes boundary information corresponding to clearly defined boundary segments and boundary information corresponding to unclear boundary segments. Accordingly, based on the planned movement mode, when controlling the autonomous working machine 10 to move and work within the work area, a third sub-region can be determined in the work area first based on the boundary information corresponding to the clearly defined boundary segments, and then the autonomous working machine 10 can be controlled to work in the third sub-region, wherein at least one edge in the third sub-region is a clearly defined boundary segment.

[0312] For example Figure 8aAs shown, after the mapping is completed, the first map of the boundary of the working area A is obtained. Segment a8-b8 and segment c8-d8 are clear boundary segments. We can first perform "bow-shaped cutting" (the cutting path is indicated by black arrows) in the third sub-region formed by segment a8-b8 and segment c8-d8. No feature acquisition is performed during the cutting process.

[0313] After the work in the third sub-region is completed, the machine can move along the unclear boundary section and perform feature re-sampling. After moving to the clear boundary section, the feature re-sampling ends, and this unclear boundary section where feature re-sampling was performed also becomes the clear boundary section. Based on this clear boundary section, a new third sub-region can be determined, and then the autonomous machine 10 is controlled to work in this new third sub-region.

[0314] For example Figure 8b As shown, after the "bow-shaped cut" is completed in the third sub-region formed by segments a8-b8 and c8-d8, the system moves along the unclear boundary segment d8-b8 and performs feature acquisition. The orange arrows in the figure represent the feature acquisition path, and the circles represent the feature acquisition positions, i.e., the target positions. After moving to position b8, the system performs a "bow-shaped cut" (the cutting path is indicated by green arrows) in the new third sub-region corresponding to segment d8-b8. No feature acquisition is performed during the cutting process.

[0315] If there are still unclear boundary segments in the first map, continue moving along the unclear boundary segments and perform feature acquisition, and so on, until there are no more unclear boundary segments or until the work in the entire working area is completed.

[0316] In this approach, work is first performed on the third sub-region corresponding to the clearly defined boundary segments. This means work begins immediately after mapping is completed, reducing the user's waiting time for the autonomous mobile machine to start operating. From a user experience perspective, this shortens the waiting time between the creation of the first map and the start of the autonomous mobile machine's operation, thus improving the user experience. By supplementing features along the boundary segments with unclear boundaries, the robustness of the autonomous mobile machine's positioning near the boundary is enhanced, making it less likely for the autonomous mobile machine to lose its location when approaching the boundary.

[0317] Scheme Implementation Example 6

[0318] This solution adopts the following runtime sequence: "Map building → Feature acquisition during operation → Autonomous working machine 10 operation".

[0319] After obtaining the first map, the autonomous working machine 10 can be controlled to move and work within the working area based on the planned movement mode, according to the first map. The working area is a part of the working area.

[0320] The first map mentioned above includes boundary information corresponding to clearly defined boundary segments. Accordingly, based on the planned movement mode, when controlling the autonomous machine 10 to move and work within the work area, a third sub-region can be determined within the work area first based on the boundary information corresponding to the clearly defined boundary segments. Then, the autonomous machine 10 is controlled to work within this third sub-region, where at least one edge is a clearly defined boundary segment. Feature acquisition is performed when turning at the boundary during the work process; this is essentially feature acquisition at the clearly defined boundary location, thus enriching the visual features at the clearly defined boundary location. Feature acquisition is also performed at the target location, which is located on or within a preset distance from the clearly defined boundary segment.

[0321] For example Figure 9 As shown, after mapping is completed, a first map of the boundary of work area A is obtained. Segments a9-b9 and c9-d9 are clear boundary segments. A "bow-shaped cut" (the cutting path is indicated by black arrows) can be performed first in the third sub-region formed by segments a9-b9 and c9-d9. During the cutting process, cutting operations are performed within the third sub-region. For example, cutting is performed from the boundary position towards the interior of the third sub-region and through its interior. After reaching the opposite boundary (or near the boundary, such as within 0.2 meters), the cutting direction is turned and moved along the boundary for a distance (e.g., 0.4 meters), then the cutting direction is turned towards the interior of the third sub-region and through its interior, and so on, thus completing the mowing work in the work area. During the work, the coordinate information of the boundary position in the first map and / or visual features are used to determine whether the boundary has been reached and to move along the boundary.

[0322] Unlike embodiment five, feature acquisition is performed when turning at the boundary during the operation. Figure 9 The purple arrow indicates when feature acquisition is performed during a turn. After completing the work and feature acquisition in the third sub-region, movement and feature acquisition along the unclear boundary ceases, and work can begin within the entire work area A; or, after completing the work and feature acquisition in the third sub-region, movement and feature acquisition along the unclear boundary are performed, and then work can begin within the entire work area A.

[0323] This approach begins by working on the third sub-region corresponding to clearly defined boundary segments. This means that work commences immediately after mapping is complete, reducing the user's waiting time for the autonomous mobile machine to begin operation. From a user experience perspective, this shortens the time between map creation and the autonomous mobile machine's operation, thus improving the overall experience. Furthermore, feature acquisition is performed when turning at clearly defined boundary locations during the process, resulting in richer visual features at those locations. The entire process avoids movement along the boundaries and feature acquisition, making it easier for users to understand the autonomous mobile machine's actions.

[0324] Scheme Example 7

[0325] This solution adopts a runtime sequence of "map creation → feature acquisition while working".

[0326] After acquiring the first map, the autonomous working machine 10 can be controlled to move and work within the working area based on the first map. Specifically, at least two target boundary segments can be determined based on the first map, and the autonomous working machine 10 can be controlled to move along the target boundary segments. Among them, a semi-random movement mode can be used between the target boundary segments, controlling the autonomous working machine 10 to move through the interior of the working area from one target boundary segment to another, performing work during the movement, and performing feature supplementation when turning at the boundary.

[0327] In this scheme, the aforementioned target boundary segment can be a clear boundary segment or an unclear boundary segment.

[0328] In this case, if the autonomous working machine 10 is to be controlled to move from one target boundary segment to another target boundary segment within the working area, a reference position can be determined in the other target segment, and the distance between the reference position and the previous target position in the other target segment is greater than a first threshold.

[0329] For example Figure 10As shown in the diagram, after mapping is completed, a first map of the boundary of work area A is obtained. Segments a10-b10, c10-d10, and e10-f10 on the boundary are clear boundary segments, represented by red lines in the diagram. Assuming the initial mapping position is position g10 on segment a10-b10, the robot can move a predetermined distance (e.g., 0.2 meters) along the mapping direction, then turn and move towards one of the clear boundary segments (let's say cd) within work area A. Since segment c10-d10 is the first clear boundary segment moved to, the robot can semi-randomly move to position h10 on segment c10-d10, with feature acquisition occurring during boundary turning. In the diagram, black arrows indicate the movement direction of the automatic lawnmower, and purple arrows indicate feature acquisition during boundary turning. After moving to position h10, the robot moves a predetermined distance along the mapping direction, and the reached point is taken as the target position. For example, move a distance (e.g., 0.2 meters) along the mapping direction to reach position point i10 in segment c10-d10. At position point i10, turn to perform feature acquisition. Then, within the working area A, move towards another clear boundary segment (let's say segment e10-f10) and work, again moving randomly or semi-randomly to position point l10 in segment e10-f10. After reaching position point l10, move 0.2 meters along the mapping direction to reach position point m10 in segment e10-f10. At position point m, turn to perform feature acquisition. Then, within the working area A, move towards another clear boundary segment (let's say a10-b10) and work, semi-randomly moving to position point j10 in segment a10-b10. After reaching position j10, move 0.2 meters along the mapping direction to position k10 in segment a10-b10. At position k10, turn to perform feature acquisition. Then move towards another clear boundary segment (let's say segment e10-f10) and work, semi-randomly moving to position n10 in segment e10-f10. However, it is necessary to ensure that the distance between position n10 (i.e., the reference point) and the previous target position (i.e., position m10) in segment e10-f10 is greater than a first threshold. This first threshold is usually a value on the order of meters, such as 2 meters or 3 meters. By continuously executing the above process, the features of the clear boundary are gradually acquired.

[0330] After the features of the clear boundary are fully acquired, for example, by moving to the endpoint of the clear boundary (i.e., the starting position of the clear boundary), or by moving to a preset distance range from the clear boundary (e.g., within 0.2 meters), the unclear boundary segment can be used as the target boundary segment. Using a similar method, a semi-random movement mode is employed between the unclear boundaries to control the automatic lawnmower to move and work within the work area A. This process is the same as the process of using the clear boundary segment as the target boundary segment, and will not be elaborated here. Through this process, the features of the unclear boundary are gradually acquired, thus achieving feature acquisition of the entire work area boundary.

[0331] In this approach, the system first moves semi-randomly within the working area based on clearly defined boundary segments, and then moves semi-randomly between segments with unclear boundary segments. Furthermore, feature acquisition is performed during boundary turns. On one hand, this allows for immediate work after map creation, reducing the user's waiting time for the autonomous machine to begin operation. This shortens the time from map creation to the autonomous machine's operation, improving the user experience. On the other hand, simultaneous feature acquisition during operation results in richer visual features at the boundary locations.

[0332] Scheme Implementation Example 8

[0333] This solution adopts a runtime sequence of "map creation → feature acquisition while working".

[0334] After acquiring the first map, the autonomous working machine 10 can be controlled to move and work within the working area based on the first map. Specifically, at least two target boundary segments can be determined based on the first map, and the autonomous working machine 10 can be controlled to move along the target boundary segments. Among them, a semi-random movement mode can be used between the target boundary segments, controlling the autonomous working machine 10 to move through the interior of the working area from one target boundary segment to another, performing work during the movement, and performing feature supplementation when turning at the boundary.

[0335] In this scheme, the aforementioned target boundary segment can be a part of the boundary of the working area, that is, moving from one part of the boundary of the working area towards another part of the working area (usually the opposite part). The starting position of the target boundary segment can be the starting position of mapping, or a boundary position where the positioning accuracy meets the visual loop closure condition (the meaning of the visual loop closure condition can be found in the relevant description in Scheme Implementation Example 4), or the starting position of a clear boundary segment, or other boundary positions, etc.

[0336] In this case, if the autonomous working machine 10 is to be controlled to move from one target boundary segment to another target boundary segment within the working area, a reference position can be determined in the other target boundary segment, which is the previous target position in the other target boundary segment.

[0337] by Figure 11 As shown, after mapping is completed, a first map of the boundary of work area A is obtained. Based on this first map, two target boundary segments are determined, and these two segments are connected to form the entire boundary. A starting position is determined on the boundary; this starting position is taken as the mapping starting position. Assuming the starting position is p1, the automatic lawnmower moves a preset distance (e.g., 0.4 meters) along the mapping direction from p1, then turns and moves towards the opposite boundary segment within work area A, performing work (i.e., cutting), until it reaches position p2. Feature acquisition is performed during the boundary turning. In the diagram, black arrows indicate the movement direction of the automatic lawnmower, and purple arrows indicate feature acquisition during boundary turning. After reaching position p2, it moves 0.4 meters along the mapping direction to p3, then turns and performs feature acquisition, moving towards position p1 within work area A and cutting. After reaching position p1, it moves 0.4 meters along the mapping direction to p4, then turns and performs feature acquisition, moving towards position p3 within work area A and cutting. After reaching position p3, move 0.4 meters along the mapping direction to reach position p5, then turn and perform feature acquisition. Within the working area A, move towards position p4 and cut. After reaching position p4, move 0.4 meters along the mapping direction to reach position p6, then turn and perform feature acquisition again. Within the working area A, move towards position p5 and cut. Continue this process to complete feature acquisition of the entire working area while cutting.

[0338] In this approach, after mapping is completed, the autonomous machine moves semi-randomly within the work area and performs its tasks, while simultaneously acquiring additional features when turning at boundaries. This reduces the user's waiting time for the autonomous machine to begin working, thus shortening the time between map creation and the machine's operation and improving the user experience. Furthermore, the on-the-spot feature acquisition during operation results in richer visual features at boundary locations. Since no movement or feature acquisition is performed along the boundaries, the autonomous machine's actions are easier for users to understand.

[0339] Scheme Example 9

[0340] This solution adopts a runtime sequence of "map creation → feature acquisition while working".

[0341] After acquiring the first map, the autonomous working machine 10 can be controlled to move and work within the working area based on the first map. Specifically, at least two target boundary segments can be determined based on the first map, and the autonomous working machine 10 can be controlled to move along the target boundary segments. Among them, a semi-random movement mode can be adopted between the target boundary segments, controlling the autonomous working machine 10 to move through the interior of the working area from one target boundary segment to another, performing work during the movement, and performing feature supplementation when turning at the boundary.

[0342] In this scheme, the aforementioned target boundary segment can be part of the boundary of the work area, that is, moving from one part of the boundary of the work area towards another part of the work area (usually the opposite part). The starting position of the target boundary segment can be located in the clear boundary segment, that is, a position on the clear boundary segment.

[0343] In this case, if the autonomous working machine 10 is to be controlled to move from one target boundary segment to another target boundary segment within the working area, a reference position can be determined in the other target segment. The reference position can be located between adjacent historical target positions.

[0344] by Figure 12a For example, after mapping is completed, a first map of the boundary of working area A is obtained, with red line segments representing clear boundary segments. In this embodiment, two clear boundary segments can be selected as target boundary segments. The angle between the two selected clear boundary segments can meet preset angle requirements, for example, the angle between the two clear boundary segments is the maximum, or the angle between the two clear boundary segments is greater than a preset angle threshold.

[0345] The angle between two clearly defined boundary segments can be represented by the angle between the line segments formed by the midpoints of the clearly defined boundary segments and the center point of the working area. For example... Figure 12b For example, the angle between sharp boundaries l1 and l2 is α1, the angle between sharp boundaries l1 and l3 is α2, and the angle between sharp boundaries l2 and l3 is α3. Among them, α2 is the largest, so sharp boundary segments l1 and l3 can be selected as the two target boundary segments.

[0346] From a location on one of the target boundary segments (let's assume it's...) Figure 12aStarting from the boundary position (s12) in the diagram, the automatic lawnmower moves along the boundary. The movement process is represented by black arrows in the diagram. When relocation occurs, the position p1 is recorded. In this embodiment, "relocation" means at least satisfying the visual loop closure condition (the specific meaning of the visual loop closure condition can be found in the relevant description in embodiment four), and may further include being able to determine the current pose of the automatic lawnmower based on the visual features of the image. Feature acquisition is performed at position p1, i.e., position p1 is the target position point, and feature acquisition is represented by purple arrows in the diagram. The automatic lawnmower continues to move along the boundary from position p1. When it moves to position p2, relocation occurs and the distance between p2 and p1 is greater than a preset threshold. Then, position p2 is recorded, feature acquisition is performed at position p2, and the automatic lawnmower is controlled to move towards another target boundary segment within the working area A, starting from position p2, using a semi-random movement mode. Assuming it moves to position a12, the automatic lawnmower moves and cuts within the working area A.

[0347] Starting from position a12, the robot moves along the boundary. When relocation occurs, position q1 is recorded, and feature acquisition is performed at position q1, which is the target location. Continuing to move along the boundary from position q1, when relocation occurs at position q2 and the distance between q2 and q1 is greater than a preset threshold, position q2 is recorded, and feature acquisition is performed at position q2. The robot then switches to a semi-random movement mode starting from position q2, controlling the automatic lawnmower to move within the work area A towards another target boundary segment, moving to the point between the last two target positions, p1 and p2, for example, the midpoint b12. After reaching position b12, the robot moves along the boundary from position b12. When relocation occurs at position p3 and the distance between p3 and p2 is greater than a preset threshold, feature acquisition is performed at position p3, and the robot switches to a semi-random movement mode starting from position p3, controlling the lawnmower to move within the work area A towards another target boundary segment, to the point between q1 and q2, for example, the center point c12. After reaching position c12, it starts moving along the boundary from position c12. When relocation occurs at position q3 and the distance between q3 and q2 is greater than a preset threshold, feature supplementation is performed at position q3, and the machine turns to adopt a semi-random movement mode starting from position q3, controlling the lawnmower to move to another target boundary section within the working area A between q2 and q3, and so on, until the stopping condition is met.

[0348] The above process actually records a series of target positions: p1-p2-q1-q2-p3-q3-p4-q4-… If the distance between the last target position and the first target position is less than a preset threshold, it indicates that feature acquisition between the two target boundary segments has been completed, and the stopping condition can be considered met. At this time, if there are still segments on the boundary of working area A that have been moved to for feature acquisition, feature acquisition can be performed directly along the edge of these segments.

[0349] Furthermore, when controlling the autonomous machine to move within the work area using a semi-random movement mode, abnormal situations may arise, such as the detection of a prohibited area. This prohibited area could be an area containing obstacles, etc. In response to the autonomous machine 10 detecting a prohibited area, it stops moving to the predetermined reference position and returns to the previous target boundary segment. Figure 12c For example, assuming that during the process described above, when moving from position p2 within the working area A to another target boundary segment, a prohibited area is detected, the system can return to the previous target boundary segment, for example, moving to position b12 between p1 and p2. Then, starting from position b12, the system moves along the boundary. When relocation occurs at position p3 and the distance between p3 and p2 is greater than a preset threshold, feature acquisition is performed at position p3, and the system switches to a semi-random movement mode starting from position p3, controlling the lawnmower to move within the working area A to another target boundary segment.

[0350] In this approach, after mapping is completed, the system moves semi-randomly within the work area and performs tasks, supplementing feature collection when turning at boundaries during the process. This reduces the user's waiting time for the autonomous machine to begin work, shortening the time between map creation and the machine's operation, thus improving the user experience. Furthermore, simultaneous feature collection during operation results in richer visual features at boundary locations. It also effectively addresses anomalies such as the presence of restricted areas within the work area.

[0351] In Examples 7 to 9, before the autonomous working machine 10 moves within the working area based on a semi-random movement pattern and performs feature supplementation, it can also move along the boundary and perform feature supplementation first.

[0352] Scheme Example 10

[0353] This solution adopts a runtime sequence of "map creation → feature acquisition along the boundary → feature acquisition while working".

[0354] After acquiring the first map, the autonomous working machine 10 can be controlled to move along the boundary at least one lap based on the first map. During the movement, in response to the first information satisfying the first condition and the machine position being located in an unclear boundary segment, the autonomous working machine 10 is controlled to perform feature supplementation at that position, thereby achieving feature supplementation along the boundary. The first information satisfying the first condition includes: at the current machine position, the autonomous working machine 10 can successfully detect visual loop closure; and the machine position is located on or near the boundary.

[0355] After moving around the boundary, the autonomous working machine 10 is controlled to perform feature acquisition while working within the working area. Multiple reversal positions can be determined on the boundary of the working area based on the first map. The autonomous working machine 10 is controlled to move and work within the working area in a semi-random movement mode. In the semi-random movement mode, the autonomous working machine 10 moves towards the reversal position, turns at the reversal position, and moves to other reversal positions. The autonomous working machine 10 performs feature acquisition by turning at the reversal position, moves to other reversal positions, and turns again, and so on, thereby realizing feature acquisition while working.

[0356] The autonomous working machine 10 can avoid going out of bounds by turning when it encounters a reversing position. Furthermore, multiple reversing positions can be evenly distributed on the boundary.

[0357] The autonomous working machine 10 performs feature acquisition at reversing positions where visual loop relationships can be successfully detected; if visual loop relationships are not successfully detected at visual loop positions, feature acquisition is not required, and the machine simply turns and moves to other reversing positions.

[0358] by Figure 13a As shown, the boundary of the first map includes clear boundary segments and unclear boundary segments. The unclear boundary segments include segments a13-b13, c13-d13, and e13-f13, while the clear boundary segments include segments f13-a13, b13-c13, and d13-e13. After obtaining the first map, the autonomous working machine 10 moves along the boundary of the working area according to the first map. It can move along the boundary in a counterclockwise direction, as shown in the diagram. Figure 13a As indicated by the orange arrow, no feature acquisition is performed when moving to a clear boundary segment; feature acquisition is performed when moving to an unclear boundary segment, for example, in... Figure 13a Feature sampling is performed at the location marked by the middle circle. This allows for targeted feature sampling of unclear boundary sections, updating the initial map and improving the machine's positioning accuracy near unclear boundaries.

[0359] Moving around the boundary allows for the acquisition of features for all unclear boundaries, and then multiple reversal positions are determined based on the updated first map. Figure 13b As shown in the diagram, triangles represent various reversal positions. Starting from any position in the working area, such as position q1, the autonomous machine 10 moves to any reversal position, for example, to reversal position q2. At reversal position q2, it turns right at an angle β2 and moves towards reversal position q3. At reversal position q3, it turns right at an angle β3 and moves towards reversal position q4. At reversal position q4, it turns right at an angle β4 and moves towards reversal position q5. During the turning at reversal positions q2, q3, and q4, the first updated image is acquired during the turning process, and feature acquisition is performed again to update the first map a second time.

[0360] Specifically, the turning angle can be randomly determined within a preset angle range based on the machine's current position, current direction of movement, and the relative positional relationship with other reversing positions. For example, if the preset angle range is 120 to 160 degrees, after the autonomous machine 10 reaches reversing position q4, it determines a first direction based on its current position and current direction of movement. Then, it randomly selects an angle within the preset angle range, such as 150 degrees, and designates the direction forming a 150-degree angle with the first direction as the second direction. Other reversing positions located in the second direction (i.e., reversing position q5) are then identified as candidate positions. The autonomous machine 10 is then controlled to turn from reversing position q4 and move towards reversing position q5. If the randomly selected angle within the preset angle range is 140 degrees, then position q6 is a candidate position. If the randomly selected angle within the preset angle range is 160 degrees, then position q7 is a candidate position.

[0361] As a further implementation, in the mapping and feature acquisition process of the above-described embodiments, a second map can be generated using the first map. The second map includes coordinate information and / or visual information of multiple locations on the boundary, such as images or visual features extracted from images, or coordinate information and / or visual information of multiple locations within a preset distance from the boundary. The second map is then sent to the user terminal so that the user terminal can display the second map.

[0362] The second map displayed on the user terminal clearly shows the coordinates and / or visual features of multiple locations on or within a preset distance from the boundary of the work area. Since there is a correspondence between the first and second maps—for example, a correspondence between location coordinates and between location coordinates and visual features—the user can use the second map to set boundary attributes and adjust boundary positions on the first map.

[0363] For example, the interface displaying the second map can show page elements for setting boundary attributes. Users can trigger these page elements to set boundary attributes on the second map. In response to the user's boundary attribute settings on the second map, the user terminal sends boundary attribute information to the autonomous machine 10. The autonomous machine 10 then sets the boundary attributes based on this information. The set boundary attributes can include clearly defined and unclear boundary sections from the first map, and can also include attributes such as safety, danger, slope, no-entry, and driving path (driving only, no mowing), etc.

[0364] For example, a user can adjust the boundary position of the work area on the second map by triggering page elements or using specific gestures (such as clicking or dragging). In response to the user's boundary position adjustment on the second map, the user terminal sends boundary position adjustment information to the autonomous working machine 10. The autonomous working machine 10 then adjusts the position of the corresponding boundary on the first map based on this boundary position adjustment information.

[0365] Through this embodiment, users can flexibly edit the first map by setting boundary attributes and adjusting its position, and supplement and correct the first map in a timely manner, thereby improving the accuracy of the first map.

[0366] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0367] Furthermore, this application also provides an embodiment of a control device applied to an autonomous working machine 10. The autonomous working machine 10 is configured to move and / or work in a work area. The autonomous working machine 10 is equipped with a camera. The control device includes: a map acquisition unit 1401, a motion control unit 1402, and a feature acquisition unit 1403. It may further include a map synchronization unit 1404 and an attribute configuration unit 1405. The main functions of each component are as follows:

[0368] Map acquisition unit 1401 is configured to acquire a first map of the boundaries of the work area.

[0369] The movement control unit 1402 is configured to control the movement of the autonomous working machine 10 according to the first map.

[0370] The feature acquisition unit 1403 is configured to acquire features of the first map during movement. The feature acquisition includes: in response to the autonomous working machine 10 moving to each target location, acquiring a first updated image at the target location, wherein the target location is located on the boundary or within a preset distance from the boundary, and the first updated image represents the image corresponding to the surrounding environment of the boundary; and updating the first map based on the first updated image.

[0371] As a first feasible approach, the motion control unit 1402 can be specifically configured to control the autonomous working machine 10 to move along the boundary according to the first map.

[0372] As a second possible approach, the motion control unit 1402 can be specifically configured to control the autonomous working machine 10 to move and work within the working area according to the first map.

[0373] In some embodiments of this application, the feature acquisition unit 1403 may also be configured to terminate feature acquisition in response to the autonomous working machine 10 returning to the starting position of the movement, or the distance between the machine position of the autonomous working machine 10 and the starting position being less than a distance threshold.

[0374] In the first achievable method described above, the mobile control unit 1402 can also be configured to: determine the boundary segment traversed by the feature acquisition; control the autonomous working machine 10 to work in the first sub-region in response to the boundary segment forming a first sub-region; and return to execute, according to the first map, control the autonomous working machine 10 to move along the boundary and perform subsequent processing until the autonomous working machine 10 returns to the starting position of the movement.

[0375] In the first feasible method described above, the first map may include boundary information corresponding to multiple boundary segments. The motion control unit 1402 may be specifically configured to: determine a second sub-region based on the boundary information corresponding to the multiple boundary segments included in the first map; control the autonomous working machine 10 to move along the boundary segments of the second sub-region; and may also be configured to: control the autonomous working machine 10 to move and work in the second sub-region in response to the feature acquisition unit 1403 completing feature acquisition of the boundary segments of the second sub-region.

[0376] In the first feasible approach described above, the first map may include boundary information corresponding to unclear boundary segments, wherein when the autonomous machine 10 is located in an unclear boundary segment, the autonomous machine 10 cannot identify the boundary position based solely on the image. The motion control unit 1402 may be specifically configured to control the autonomous machine 10 to move along the unclear boundary segment according to the first map.

[0377] Specifically, when the motion control unit 1402 controls the autonomous working machine 10 to move along an unclear boundary section according to the first map, it can be configured to start from the boundary position where the positioning accuracy meets the visual loop closure condition, and control the autonomous working machine 10 to move along the unclear boundary section according to the first map.

[0378] In the second possible implementation method described above, the motion control unit 1402 can be specifically configured to: control the autonomous working machine 10 to move and work in the working area based on the planned movement pattern, wherein the working area is a part or all of the working area.

[0379] In different rounds of global work tasks, the working path of the autonomous working machine 10 is different. The global work task is based on the planned movement mode, which controls the autonomous working machine 10 to move and work within the work area.

[0380] In some embodiments of this application, the first map may include boundary information corresponding to clear boundary segments. When the autonomous machine 10 is located in a clear boundary segment, the autonomous machine 10 can identify the boundary position based solely on the image. When the motion control unit 1402 controls the autonomous machine 10 to move and work within the work area based on a planned movement mode, it can be specifically configured to: determine a third sub-region within the work area based on the boundary information corresponding to the clear boundary segments, wherein at least one edge of the third sub-region is a clear boundary segment; and control the autonomous machine 10 to move and work within the third sub-region based on the planned movement mode.

[0381] In some embodiments of this application, the first map also includes boundary information corresponding to clear boundary segments. When the autonomous machine 10 is located in a clear boundary segment, the autonomous machine 10 can identify the boundary position based solely on the image. In this case, the target position adopted by the feature acquisition unit 1403 is located on or within a preset distance from the clear boundary segment.

[0382] In the second possible implementation described above, the motion control unit 1402 can be specifically configured to: determine at least two target boundary segments based on the first map, and control the autonomous working machine 10 to move along the target boundary segments. A semi-random movement mode is used between the target boundary segments, controlling the autonomous working machine 10 to traverse the interior of the work area to move from one target boundary segment to another.

[0383] Among them, the target boundary segment is a clear boundary segment. When the autonomous working machine 10 is located in the clear boundary segment, the autonomous working machine 10 can identify the boundary position based solely on the image.

[0384] Alternatively, the target boundary segment is an unclear boundary segment. When the autonomous machine 10 is located in an unclear boundary segment, the autonomous machine 10 cannot identify the boundary position based solely on the image.

[0385] Alternatively, the target boundary segment is part of the boundary of the working area, and the positioning accuracy of the starting position in the target boundary segment meets the visual loop closure condition;

[0386] Alternatively, the target boundary segment is part of the boundary of the working area, and the starting position of the target boundary segment is located in the clear boundary segment.

[0387] In some embodiments of this application, when the motion control unit 1402 controls the autonomous working machine 10 to move from one target boundary segment to another by traversing the interior of the working area in a semi-random movement mode, it can determine a reference position in the other target boundary segment. The distance between the reference position and the previous target position in the other target boundary segment is greater than a first threshold, or the reference position is the starting position of the other target boundary segment, or the reference position is the previous target position in the other target boundary segment, or the reference position is located between adjacent historical target positions. Then, based on the semi-random movement mode, the autonomous working machine 10 is controlled to traverse the interior of the working area and move to the reference position.

[0388] If the reference position is located between adjacent historical target positions, the movement control unit 1402, in the process of controlling the autonomous working machine 10 to pass through the interior of the working area based on the semi-random movement mode, controls the autonomous working machine 10 to stop moving to the reference position and return to the target boundary segment in response to the autonomous working machine 10 detecting a prohibited area.

[0389] In some embodiments of this application, the motion control unit 1402 may be specifically configured to: determine multiple reversing positions on the boundary of the work area according to the first map; control the autonomous working machine 10 to move and work in the work area in a semi-random movement mode, wherein in the semi-random movement mode, the autonomous working machine 10 moves toward the reversing position, turns at the reversing position and moves toward other reversing positions.

[0390] In each of the above-mentioned possible methods, during the movement of the automatic working machine, the feature acquisition unit 1403 determines the machine position of the autonomous working machine 10 and the first information corresponding to the machine position. The first information includes at least one of the following parameters: visual positioning accuracy, distance between the machine position and the previous target position, and time interval between the current time point and the time point when the machine moves to the previous target position. In response to the first information satisfying the first condition, the autonomous working machine 10 is determined to move to the target position.

[0391] In some embodiments of this application, the first map may include boundary information corresponding to unclear boundary segments, wherein when the autonomous machine 10 is located in an unclear boundary segment, the autonomous machine 10 cannot identify the boundary position based solely on the image.

[0392] In response to the first information satisfying the first condition and the machine position being located in an unclear boundary section, the feature acquisition unit 1403 determines that the autonomous working machine 10 has moved to the target position.

[0393] Among them, when the feature acquisition unit 1403 determines that the autonomous working machine 10 has moved to the target position in response to the first information satisfying the first condition, it can be specifically configured to: determine the machine position as the target position in response to the machine position being located on or near the boundary and the positioning accuracy of the machine position meeting the visual loop closure condition.

[0394] In the second possible implementation method described above, the feature acquisition unit 1403 is configured to acquire a first updated image through the camera of the autonomous machine 10 at the target position where the autonomous machine 10 turns based on the first map.

[0395] Among the above-mentioned feasible methods, when the feature acquisition unit 1403 acquires the first updated image at the target location, it may employ, but is not limited to, the following methods:

[0396] The first method involves controlling the autonomous working machine 10 to rotate at the target position, and during the rotation, acquiring a first updated image based on the camera of the autonomous working machine 10. The first updated image includes at least two images from different angles.

[0397] The second method involves controlling the camera to rotate relative to the body of the autonomous working machine 10, and during the rotation of the camera, acquiring a first updated image based on the camera of the autonomous working machine 10. The first updated image includes at least two images from different angles.

[0398] The third method: control the panoramic camera of the autonomous working machine 10 to collect panoramic images, and the first updated image is a panoramic image.

[0399] In each of the above-described possible implementations, the map acquisition unit 1401 can be specifically configured to: control the autonomous machine 10 to move along the boundary and perform mapping, the mapping including: acquiring inertial navigation information from the inertial navigation unit of the autonomous machine 10 and a first initial image captured by the camera of the autonomous machine 10, wherein the shooting angle corresponding to the first initial image is different from the shooting angle corresponding to the first updated image; generating a first map using the inertial navigation information and visual features extracted from the first initial image; or, receiving the first map from a server or user terminal; or, receiving a map download instruction from a server or user terminal and downloading the first map according to the map download instruction.

[0400] As one of the possible ways to achieve this, the autonomous working machine 10 moves in the same or opposite directions during the mapping process and the feature acquisition process.

[0401] Furthermore, the mapping process also includes: acquiring a second updated image at the mapping start position on the boundary, the second updated image representing the image of the surrounding environment at the mapping start position; and ending the mapping process in response to the autonomous working machine 10 returning to the mapping start position.

[0402] In the above-mentioned possible implementation methods, when the feature acquisition unit 1403 updates the first map according to the first updated image, it can be specifically configured to: acquire the inertial navigation information corresponding to the first updated image; determine the feature points of the first updated image; determine the descriptive information of the feature points, the position information of the feature points in the image, and the three-dimensional coordinate information of the feature points; and update the first map according to the inertial navigation information, the descriptive information of the feature points, the position information of the feature points in the image, and the three-dimensional coordinate information of the feature points.

[0403] In each of the above-mentioned possible implementation methods, the map synchronization unit 1404 is configured to generate a second map using the first map. The second map includes coordinate information and visual features of multiple locations on the boundary, and coordinate information and visual features of multiple locations within a preset distance range from the boundary. The second map is then sent to the user terminal so that the user terminal can display the second map.

[0404] In each of the above-mentioned possible implementation methods, the attribute configuration unit 1405 is configured to receive boundary attribute information sent by the user terminal; and determine the clear boundary segments and unclear boundary segments in the first map based on the boundary attribute information.

[0405] The control device can be a control chip.

[0406] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention.

[0407] Based on the same inventive concept as the foregoing embodiments, this embodiment of the invention provides an autonomous working machine 10, wherein the width of the autonomous working machine 10 is greater than or equal to 60 centimeters, such as... Figure 15 As shown, the autonomous working machine 10 includes: a processor 1510 and a memory 1511 storing computer programs; wherein, Figure 15The processor 1510 shown in the diagram does not indicate that there is only one processor 1510, but only indicates the positional relationship of processor 1510 relative to other devices. In practical applications, there can be one or more processors 1510; similarly, Figure 15 The memory 1511 illustrated in the diagram has the same meaning, that is, it is only used to indicate the positional relationship of memory 1511 relative to other devices. In practical applications, there can be one or more memories 1511. When the processor 1510 runs the computer program, it implements the control method described in the above embodiments.

[0408] The autonomous machine 10 may further include at least one network interface 1512. The various components of the autonomous machine 10 are coupled together via a bus system 1513. It is understood that the bus system 1513 is used to enable communication between these components. In addition to a data bus, the bus system 1513 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in… Figure 15 The general designated all buses as Bus System 1513.

[0409] The memory 1511 can be volatile memory or non-volatile memory, or both. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), ferromagnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD-ROM); the magnetic surface memory can be disk storage or magnetic tape storage. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Synchronous Static Random Access Memory (SSRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), SyncLink Dynamic Random Access Memory (SLDRAM), and Direct Rambus Random Access Memory (DRRAM).The memory 1511 described in some embodiments of this application is intended to include, but is not limited to, these and any other suitable types of memory.

[0410] In some embodiments of this application, the memory 1511 is used to store various types of data to support the operation of the autonomous working machine 10. Examples of this data include: any computer programs used to operate on the autonomous working machine 10, such as operating systems and applications; contact data; phonebook data; messages; pictures; videos, etc. The operating system includes various system programs, such as a framework layer, core library layer, driver layer, etc., used to implement various basic services and handle hardware-based tasks. Applications may include various applications, such as media players, browsers, etc., used to implement various application services. Here, programs implementing the methods of some embodiments of this application may be included in the application.

[0411] Based on the same inventive concept as the foregoing embodiments, this embodiment also provides a computer storage medium storing a computer program. The computer storage medium can be a magnetic random access memory (FRAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM), etc.; it can also be various devices including one or any combination of the above-mentioned memories, such as mobile phones, computers, tablet devices, personal digital assistants, etc. When the computer program stored in the computer storage medium is executed by a processor, it implements the control method applied to the aforementioned autonomous working machine 10. The specific steps implemented when the computer program is executed by the processor are described in the method embodiment and will not be repeated here.

[0412] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0413] In this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, which includes not only the elements listed but also other elements not expressly listed.

[0414] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A control method for an autonomous working machine (10), applied to said autonomous working machine (10), characterized in that, The control method includes: Obtain the first map of the boundaries of the work area (A); Based on the first map, control the movement of the autonomous working machine (10); During the movement, feature acquisition is performed on the first map. Feature acquisition includes: in response to the autonomous working machine (10) moving to each target location, acquiring a first updated image at the target location, wherein the target location is located on the boundary or within a preset distance from the boundary, and the first updated image represents an image corresponding to the surrounding environment of the boundary; and updating the first map according to the first updated image. During the movement, an image of the work area (A) is acquired, and a pre-trained deep learning model is used to process the image of the work area (A) to identify the boundary between the grass area and the non-grass area, so as to control the autonomous working machine (10) to move within the work area (A).

2. The control method as described in claim 1, characterized in that, Based on the first map, controlling the movement of the autonomous working machine (10) includes: Based on the planned movement mode, the autonomous working machine (10) is controlled to move and work in the area to be worked, which is part or all of the working area (A).

3. The control method as described in claim 2, characterized in that, The autonomous working machine performs feature acquisition on the first map in at least some rounds of work tasks. The working path of the autonomous working machine (10) is different in work tasks where feature acquisition is performed and in work tasks where feature acquisition is not performed.

4. The control method as described in claim 3, characterized in that, The difference in the working path of the autonomous machine (10) between a task that performs feature acquisition and a task that does not perform feature acquisition includes at least one of the following: The path spacing of the work path is different in work tasks that perform feature acquisition and those that do not. The path direction of the work path is different in work tasks where feature supplementation is performed and in work tasks where feature supplementation is not performed. The autonomous machine (10) travels at different speeds along the work path during a task involving feature acquisition and during a task where feature acquisition is not performed.

5. The control method as described in claim 2, characterized in that, The first map includes boundary information corresponding to clear boundary segments. When the autonomous working machine (10) is located in the clear boundary segment, the autonomous working machine (10) can identify the boundary position based solely on the image. The target location is located on or within a preset distance from the clear boundary segment.

6. The control method as described in claim 1, characterized in that, Based on the first map, controlling the movement of the autonomous working machine (10) includes: Based on the first map, the autonomous working machine (10) is controlled to move along the boundary.

7. The control method according to any one of claims 1 to 6, characterized in that, The autonomous working machine (10) moves to the target location including: During the movement, the machine position of the autonomous working machine (10) and the first information corresponding to the machine position are determined. The first information includes at least one of the following parameters: visual positioning accuracy, distance between the machine position and the previous target position, and time interval between the current time point and the time point when the machine moves to the previous target position. In response to the first information satisfying the first condition, it is determined that the autonomous working machine (10) moves to the target location.

8. The control method as described in claim 7, characterized in that, The first map includes boundary information corresponding to unclear boundary segments, wherein when the autonomous working machine (10) is located in the unclear boundary segment, the autonomous working machine (10) cannot identify the boundary position based solely on the image. In response to the first information satisfying the first condition, determining that the autonomous working machine (10) has moved to the target location includes: In response to the first information satisfying the first condition and the machine location being located in the unclear boundary segment, it is determined that the autonomous working machine (10) moves to the target location.

9. The control method as described in claim 7, characterized in that, The first condition includes that the machine position is located on or near the boundary, and that the positioning accuracy of the machine position meets the visual loop closure condition; In response to the first information satisfying the first condition, determining that the autonomous working machine (10) has moved to the target location includes: In response to the machine location being located on or near the boundary, and the positioning accuracy of the machine location meeting the visual loop closure condition, the machine location is determined as the target location.

10. The control method according to any one of claims 1 to 9, characterized in that, Acquire a first updated image at the target location, including: The autonomous working machine (10) is controlled to rotate at the target position, and during the rotation, the first updated image is acquired based on the camera of the autonomous working machine (10), the first updated image including at least two images from different angles; Alternatively, the camera can be controlled to rotate relative to the body of the autonomous working machine (10), and during the rotation of the camera, the first updated image can be acquired based on the camera of the autonomous working machine (10), the first updated image including at least two images from different angles; Alternatively, the panoramic camera of the autonomous working machine (10) can be controlled to capture a panoramic image, and the first updated image is the panoramic image.

11. The control method according to any one of claims 2 to 5, characterized in that, Acquiring the first updated image at the target location includes: At the target location where the autonomous working machine (10) turns based on the first map, the first updated image is acquired by the camera of the autonomous working machine (10).

12. The control method as described in claim 11, characterized in that, At the target location where the autonomous working machine (10) turns based on the first map, the first updated image is acquired through the camera of the autonomous working machine (10), including: When the autonomous working machine (10) turns in place along the planned path based on the first map, the first updated image is captured by the camera of the autonomous working machine (10).

13. The control method as described in claim 12, characterized in that, The in-situ turning includes: Control the autonomous working machine (10) to stop moving forward; The autonomous working machine (10) is controlled to rotate around a preset rotation center by a preset angle, wherein the preset angle is less than or equal to 180 degrees.

14. The control method as described in claim 1, characterized in that, Before controlling the autonomous working machine (10) to move and perform feature acquisition according to the first map, control the autonomous working machine (10) to work; Alternatively, while controlling the autonomous working machine (10) to move and perform feature acquisition according to the first map, the autonomous working machine (10) can be controlled to work. Alternatively, after controlling the autonomous working machine (10) to move and perform feature acquisition according to the first map, the autonomous working machine (10) is controlled to work. The control of the autonomous working machine (10) to move according to the first map includes controlling the autonomous working machine (10) to move along the boundary according to the first map, or controlling the autonomous working machine (10) to move within the working area (A) according to the first map.

15. An autonomous working machine (10), characterized in that, include: Processor (1510); Memory (1511) for storing executable instructions of the processor (1510); The processor (1510) is used to perform the method according to any one of claims 1 to 14.

Images