A working area delineation method based on laser and vision solutions and an outdoor robot

Abstract

This application provides a work area delineation method and an outdoor robot based on laser and vision solutions. During the robot's movement, this application simultaneously utilizes visual and laser sensors for visual SLAM and laser SLAM recognition. Based on the visual SLAM's identification of the boundary position, the robot is triggered to obtain its accurate boundary coordinates using laser SLAM. Therefore, this application can automatically segment the grassland work boundary without prior information, building upon laser SLAM mapping of physical boundaries such as walls, and combining this with machine vision recognition of the grassland boundary. It also obtains more accurate recognition data on the coordinates of the boundary lines. This application can plan traversal paths for new grassland work areas without manual setup, achieving precise operation, overcoming the inherent limitations of visual and laser SLAM technologies, improving the positioning accuracy and precision of outdoor robots, and achieving better operational results.

Description

Technical Field

[0001] This application relates to the field of garden tools, and more specifically to a method for delineating work areas based on laser and vision schemes, and an outdoor robot. Background Technology

[0002] Existing SLAM (Simultaneous Localization and Mapping) technologies utilize cameras as visual sensors to build environmental models based on images of the external environment during movement, while simultaneously estimating the device's own position. However, in the working scenarios of automated walking devices such as lawnmowers, visual features are relatively limited. In the lawn area worked by the automated lawnmower, visual features distinguishing the grass texture can only be extracted in areas far from the lawn boundary. When the automated walking device operates within the grass, the number of identifiable feature points is scarce, leading to low accuracy in visual localization and mapping, making accurate positioning difficult. This, in turn, results in inaccurate positioning when the automated walking device follows the planned path, easily leading to walking errors (such as missed mowing or repeated mowing), and consequently, low work efficiency. Furthermore, the acquisition effect of visual sensors such as cameras is greatly affected by changes in lighting, environmental features, and texture richness. Binocular cameras or depth cameras are also costly.

[0003] When using LiDAR as a laser sensor for localization and mapping, the laser sensor has high requirements for the detectable contour length of the surrounding area. In other words, laser SLAM technology has high requirements for the detection range of the LiDAR. Furthermore, because LiDAR cannot effectively distinguish between grass and open ground, the map constructed using laser SLAM technology may include open, non-grassland surfaces.

[0004] Therefore, existing technologies often combine the aforementioned laser sensing or visual sensing technologies with the Global Positioning System (GPS) to achieve precise positioning of the equipment through real-time processing of carrier phase observations from two satellite receiving stations using Real-Time Kinematic (RTK) technology. However, satellite positioning can fail in densely wooded areas and buildings due to signal blockage. In practical applications, positioning accuracy deviations and satellite signal delays or interference often cause deviations in the equipment's operating path, affecting operational efficiency. Summary of the Invention

[0005] This application addresses the shortcomings of existing technologies by providing a work area delineation method and an outdoor robot based on laser and vision solutions. Through the collaborative operation of a visual sensing unit and a laser sensing unit, this application can accurately construct a work area boundary map and provide high-precision positioning for the outdoor robot without requiring external response signals. The specific technical solution adopted in this application is as follows.

[0006] First, to achieve the above objectives, a working area delineation method based on laser and vision solutions is proposed. The steps include: during the outdoor robot's movement, visual reference data is acquired through a visual sensing unit, and laser reference data is acquired through a laser sensing unit; when the visual reference data meets the boundary features, the boundary coordinates determined by the laser reference data at the current position are recorded; when the mapping requirements are met, the working area of ​​the outdoor robot is delineated based on the defined boundary coordinates.

[0007] Optionally, in any of the above-described laser and vision-based methods for delineating the working area, the mapping requirement is to meet any of the following conditions, or a combination thereof: during the outdoor robot's movement, the map grid corresponding to the boundary coordinates no longer increases; the boundary coordinates can be fitted to form a closed shape; the outdoor robot's movement time for determining the boundary coordinates reaches a preset time; the outdoor robot moves one lap along the boundary determined by the visual reference data.

[0008] Optionally, in any of the above-described laser and vision-based methods for defining the work area, the outdoor robot randomly selects a walking direction during its movement and performs a turning action when the visual reference data matches the boundary features.

[0009] Optionally, in any of the above-described laser and vision-based methods for defining the working area, the outdoor robot first walks to a position where the visual reference data matches the boundary features, and then performs a turning action at that position, walking around the boundary determined by the visual reference data.

[0010] Optionally, in any of the above-described laser and vision-based methods for delineating the working area, during the process of walking around the boundary determined by the visual reference data, the boundary coordinates determined by the laser reference data at the current position are recorded at preset time intervals or distance intervals.

[0011] Optionally, in any of the above-described laser and vision-based methods for defining the working area, during the outdoor robot's movement, it first moves to a position where the visual reference data matches the boundary features in any of the following ways: when the outdoor robot's base station is located on the boundary of the working area, the outdoor robot moves out of the base station and then turns backward to a position where the visual reference data matches the boundary features to the side of the base station; or the outdoor robot moves out of the base station and then first selects a direction to move in a straight line until it reaches a position where the visual reference data matches the boundary features.

[0012] Optionally, in any of the above-described laser and vision-based methods for defining the working area, the outdoor robot performs a turning action according to the following steps during its movement: based on the difference in the proportion of non-grass areas on the left and right sides of the environmental image acquired by the visual sensing unit, if the difference in proportion exceeds a preset ratio, the robot turns towards the side with a smaller proportion of non-grass areas; if the difference in proportion does not exceed a preset standard, the robot turns according to preset rules; the preset rules include any of the following turning methods: turning in a preset fixed direction, or turning randomly.

[0013] Optionally, the working area delineation method based on laser and vision solutions as described above, wherein the turning angle is set between 90° and 270°.

[0014] Meanwhile, to achieve the above objectives, this application also provides an outdoor robot, which includes: a visual sensing unit for acquiring visual reference data; a laser sensing unit for acquiring laser reference data; a first storage unit for storing computer programs or instructions; and a control unit for executing the computer programs or instructions in the storage unit, so that the outdoor robot performs the work area delineation method based on laser and vision schemes as described above.

[0015] Optionally, the outdoor robot described above may further include a second storage unit for storing the working area boundary of the outdoor robot; during the operation of the outdoor robot, the control unit is also used to compare whether the position coordinates determined by the laser reference data reach the working area boundary and / or compare whether the visual reference data conforms to the boundary features, and trigger the outdoor robot to perform a turning action when the position coordinates reach the working area boundary and / or the visual reference data conforms to the boundary features.

[0016] Beneficial effects

[0017] The work area delineation method and outdoor robot based on laser and vision solutions provided in this application utilize both visual and laser sensors on the robot for visual SLAM and laser SLAM recognition during the robot's movement. Based on visual SLAM's recognition of the robot's boundary position, laser SLAM is triggered to obtain the robot's accurate boundary coordinates. Thus, by delineating the coordinate range of each boundary during the robot's movement, this application can automatically segment the grassland work boundary without prior information. Based on laser SLAM mapping of physical boundaries such as walls, and combined with machine vision recognition of grassland boundaries, it obtains more accurate recognition data of the boundary line coordinates. This allows for precise operation by planning traversal paths for new grassland work areas without manual intervention. This application overcomes the errors caused by environmental features in visual SLAM technology, enhances the robot's positioning accuracy at the center of the work area using laser SLAM technology, and avoids misidentification of the surrounding ground during mapping, thereby improving the positioning accuracy and precision of the outdoor robot and achieving better operational results.

[0018] Other features and advantages of this application will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing this application. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate the present application and form part of the specification. Together with the embodiments of the present application, they serve to explain the present application but do not constitute a limitation thereof. In the drawings:

[0020] Figure 1 This is a flowchart of the steps of the working area delineation method based on laser and vision solutions in this application;

[0021] Figure 2 This is a top view of the working area in the first embodiment of this application;

[0022] Figure 3 This is a schematic diagram of the boundary coordinates determined based on laser reference data in the first embodiment of this application;

[0023] Figure 4 This is a schematic diagram of the map grid corresponding to the laser reference data in the first embodiment of this application;

[0024] Figure 5 This is a schematic diagram of the environment corresponding to the first turning state in the first embodiment of this application;

[0025] Figure 6 This is a schematic diagram of the environment corresponding to the second turning state in the first embodiment of this application;

[0026] Figure 7 This is a schematic diagram of the environment corresponding to the third turning state in the first embodiment of this application;

[0027] Figure 8 This is a top view of the working area in the second embodiment of this application;

[0028] Figure 9 This is a schematic diagram of the boundary coordinates determined based on laser reference data in the second embodiment of this application. Detailed Implementation

[0029] To make the objectives and technical solutions of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the described embodiments of this application without creative effort are within the scope of protection of this application.

[0030] Those skilled in the art will understand that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless defined as herein.

[0031] The meaning of "and / or" as used in this application includes situations where each exists alone or both exist simultaneously.

[0032] In this application, "inner" and "outer" refer to the direction from outside the boundary line of the working area to the inside of the boundary line of the working area, which is the inner part of the working area relative to the working area of ​​the robot itself, and vice versa; rather than a specific limitation on the device mechanism of this application.

[0033] The terms "left" and "right" as used in this application refer to the environmental image captured by the outdoor robot when the user is facing the direction in which the outdoor machine is moving, where the user's left side is considered left and the user's right side is considered right, and do not constitute a specific limitation on the device mechanism of this application.

[0034] The term "connection" as used in this application can mean a direct connection between components or an indirect connection between components through other components.

[0035] The terms "up" and "down" as used in this application refer to the direction from the track system to the bogie when the user is facing the forward direction of the bogie assembly, which is up and the direction from the bogie to the track system is down, and are not specific limitations on the device mechanism of this application.

[0036] This application provides an outdoor robot that traverses and performs tasks within a specific work area. The outdoor robot is equipped with:

[0037] The visual sensing unit is used to acquire visual reference data for visual SLAM processing.

[0038] The laser sensing unit is used to acquire laser reference data for laser SLAM processing.

[0039] The program storage unit is used to store computer programs or instructions for the control unit to call. This allows the outdoor robot's control unit to acquire visual reference data through the visual sensing unit and laser reference data through the laser sensing unit during the outdoor robot's movement and delineation of the work area. When the visual reference data matches the boundary features, the control unit records the boundary coordinates determined by the laser reference data at the current position. Finally, after obtaining the boundary coordinates that meet the mapping requirements, the control unit delineates the outdoor robot's work area based on the defined boundary coordinates.

[0040] Therefore, the outdoor robot of this application, through its onboard visual sensing SLAM unit and laser sensing SLAM unit, can automatically segment the grass work boundary without prior information. Based on laser SLAM mapping using physical boundaries such as walls, it can automatically segment the grass work boundary by recognizing the grass boundary through machine vision. Thus, after delineating the outdoor robot's work area, the outdoor robot of this application can plan a traversal path within the new grass work area based on laser positioning to perform planned lawn mowing operations.

[0041] Specifically, in the first embodiment of this application, the outdoor robot described above can be deployed in the following manner: Figure 2 Run within the indicated area to define the corresponding work area.

[0042] In this embodiment, the outdoor robot can be configured as a lawnmower with automatic walking capabilities, or other mobile work equipment with a self-propelled drive module. To power the device, a base station is typically installed near the device's work area. In non-operational mode, the device enters the base station to charge, with the base station providing shielding to prevent interference from debris. When work is required, the device leaves the base station and moves to its pre-defined work area, following a specific traversal path to perform tasks such as mowing, blowing, and vacuuming the ground within the work area.

[0043] To determine the working area of ​​the equipment and prevent it from going outside the boundary line during operation, this application can use the control unit of the automatic walking equipment to perform the following steps to obtain an accurate working area map when the equipment enters a new working area or when the working area map needs to be updated:

[0044] Step 1: The robot starts up inside the base station and executes the outbound operation program to enable the robot to move out of the station;

[0045] Step 2: After leaving the station, the robot walks randomly. During this process, the laser SLAM function and the visual VSLAM function are activated at the same time. When the robot recognizes the boundary of the work area such as grass through the visual VSLAM recognition module, it triggers the recording of the coordinate point of the current position obtained by the device through laser SLAM. This coordinate point is marked as a boundary point and recorded in the device's storage unit.

[0046] Step 3: Repeatedly drive the robot to perform Step 2 above in different random walking directions until one of the following conditions is met, at which point the machine completes mapping:

[0047] 3-a. Boundary points can be fitted as Figure 3 The closed shape indicated by the black line in the middle;

[0048] 3-b. Under laser SLAM maps, the grid no longer has similar features. Figure 4 The increment shown in the lower left corner is formed by expanding the boundary points outward from the dashed boundary;

[0049] Figure 4 Each point in the diagram represents the coordinates of the location reached during each run on the boundary line, obtained from laser SLAM. The points together form a virtual boundary line. When the device randomly turns and runs to... Figure 4 After reaching the lower left position, if visual SLAM fails to identify visual reference data matching the boundary features, the robot can continue moving forward until it finds the correct visual reference data. Figure 4 If the bottom left corner position is found to match the boundary features, the coordinates of this position can be expanded into the original list of boundary line position coordinates. The coordinates of this position can be connected with other boundary points to expand the operating area.

[0050] After leaving the station, the 3-c robot will walk randomly for a preset time, such as one or two hours, or a corresponding time calculated according to a certain proportion based on the obtained boundary area.

[0051] After leaving the station, the 3D robot can walk along the boundary determined by the visual reference data and complete a cycle through any means such as manual guidance or other recognition algorithms.

[0052] Thus, the outdoor robot can obtain a similar image by using the coordinates of each boundary point recorded in the laser SLAM mapping during its walking and traversing process, and by performing graphic fitting based on the corresponding boundary coordinates. Figure 3 The closed shape in the middle area serves as the boundary line of the robot's working area.

[0053] The aforementioned device, during step 2, can be further configured to randomly walk and confirm boundary coordinates after leaving the station: After leaving the station, it will randomly walk, and during this process, both laser SLAM and visual VSLAM functions will be activated simultaneously. This allows the robot to identify the boundary of the work area (such as grass) through the visual VSLAM recognition module, triggering the recording of the coordinates of the current position obtained by the device via laser SLAM. This coordinate point will be marked as a boundary point and recorded in the device's storage unit. Simultaneously, a turning action will be performed, adjusting the device to walk in other random directions, or according to... Figures 5 to 7 Different visual environmental features cause the robot to turn in the corresponding direction. Thus, by triggering the outdoor robot to adjust its direction of travel when it reaches the boundary position, and after a sufficiently long period of trial operation, the robot gradually accumulates boundary data of the work area.

[0054] Specifically, this application can determine the preferred turning direction by using the last recorded photo containing ground information from the visual sensing unit when the machine reaches the boundary position. For a self-propelled lawnmower robot, the ground in its working area is grass, and the grass ground in the photo belonging to the working area can be identified through features such as photo texture and pixel color range. Therefore, during machine operation, based on the machine's posture as it approaches the grass boundary, and according to the location area of ​​the grass region in the environmental image collected by the visual sensing unit, the following turning mechanism can be set:

[0055] First posture: The machine approaches from the front, and the visual sensing unit captures the data. Figure 5 The environmental image shown, after filtering and recognition processing, determines that the grassy area is located at the bottom of the photo. The machine then moves towards the grass boundary in an approximately vertical direction, approaching the boundary. For this processed image, during the computation process, the grassy area at the bottom and the non-grassy area at the top are rasterized, and the number of raster cells occupied by the blank areas (non-grass areas) above the left and right walls of the image are calculated respectively. (See below.) Figure 5As shown, when the difference in the number of grid cells on both sides does not exceed 10% of the sum of the number of grid cells on both sides, the machine can be determined to be in the first orientation facing the boundary line. When the machine reaches the boundary position, it can be triggered to rotate a preset angle in a preset direction, switching to another direction to continue running and confirm the boundary line coordinates in other directions. The preset direction in which the robot rotates during turning can be a fixed angle set manually, such as a pre-defined clockwise or counterclockwise direction, or a random direction automatically selected by the program. For example, when the device reaches the boundary line position, a direction selection program can be triggered, randomly selecting a number between 1 and 2. During turning, based on the result of the selected number, if the value is 1, the machine is controlled to turn clockwise; if the result is 2, the machine is controlled to turn counterclockwise. The specific turning angle for each turn can be set to a fixed value or randomly selected within a preset angle range of 90°-270°.

[0056] The second posture: The machine approaches at an angle, and the visual sensing unit captures the data. Figure 6 The environmental image shown has a predominantly grassy area on the lower right side. After filtering and recognition processing, the grassy area in the image can be determined to be located slightly to the right of the lower right. The camera then moves towards the grassy boundary at an angle closer to the boundary on its left side. For this processed image, during processing, the grassy area at the bottom and the non-grassy area at the top are rasterized, and the number of raster cells occupied by the blank areas (non-grass areas) above the left and right walls of the image are calculated respectively. (See below.) Figure 6 As shown, when the difference in the number of grid cells on both sides exceeds 10% of the sum of the grid cells on both sides, it can be determined that the machine is in the second type of posture, biased towards the boundary line. At this time, by comparing the size of the grid cell counts on both sides, when the machine reaches the boundary position, it can be triggered to rotate a preset angle towards the side with fewer grid cells (i.e., towards the side with more grass and farther away from the boundary), and switch to continue running in another direction to confirm the boundary line coordinates in other directions. The specific angle of rotation of the robot during turning can be a fixed angle set by the user, such as a fixed angle of 90°, 70°, etc., or an adaptive angle calculated according to a certain ratio or algorithm rule based on the difference in the number of grid cells in the non-grass areas on the left and right sides of the image. It can also be a random direction automatically selected by the program. For example, when the device reaches the boundary line position, a direction selection program can be triggered, in which a random number between 90° and 270° is randomly selected, so that when turning, the machine can be controlled to turn clockwise to the right of the forward direction by that random angle based on the result of the random number selection.

[0057] The third posture: The machine approaches at an angle, and the visual sensing unit captures the data. Figure 7The environmental image shown has a predominantly grassy area on the lower left side. After image filtering and recognition processing, the grassy area in the image can be determined to be located in the lower left of the photo. At this point, the camera moves towards the grassy boundary at an angle closer to the boundary on the right side of the camera body. For this processed image, during the processing, the grassy area at the bottom and the non-grassy area at the top of the image can be rasterized, and the number of raster cells occupied by the blank areas (non-grassy areas) above the left and right walls of the image can be calculated respectively. (See below.) Figure 7 As shown, when the difference in the number of grid cells on both sides exceeds 10% of the sum of the grid cells on both sides, it can be determined that the machine is in the third type of posture, biased towards the boundary line. At this time, by comparing the size of the grid cell counts on both sides, when the machine reaches the boundary position, it can be triggered to rotate a preset angle towards the side with fewer grid cells (i.e., towards the side with more grass and farther away from the boundary), and switch to continue running in another direction to confirm the boundary line coordinates in other directions. The specific angle of rotation of the robot during turning can be a fixed angle set by the user, such as a fixed angle of 90°, 70°, etc., or an adaptive angle calculated according to a certain ratio or algorithm rule based on the ratio of the number of grid cells in the non-grass areas on the left and right sides of the image. It can also be a random direction automatically selected by the program. For example, when the device reaches the boundary line position, a direction selection program can be triggered, in which a random number between 90° and 270° is randomly selected, so that when turning, the machine can be controlled to turn clockwise to the left of the forward direction by that random angle based on the result of the random number selection.

[0058] All of the above turning methods can prevent the equipment from running out of bounds. Furthermore, by gradually exploring the location of the boundary points of the equipment's operating range at different angles, the specific coordinates of the work boundary range can be gradually obtained.

[0059] In the second embodiment of this application, the outdoor robot can also be deployed in the following manner: Figure 8 The environment shown allows for the delineation of the corresponding working area. This implementation places higher demands on the computing power of the device processor and memory compared to the previous embodiment, and generally also places higher demands on the visual VSLAM module.

[0060] In this embodiment, when the device enters a new work area or needs to update the work area map, the following steps are performed to obtain an accurate work area map:

[0061] Step 1: The robot starts up inside the base station and executes the outbound operation program to enable the robot to move out of the station;

[0062] Step 2: After leaving the station, the robot walks randomly in the direction of the arrow. At the same time, the laser SLAM function and the visual VSLAM function are activated so that the robot first walks to a position where the visual reference data meets the boundary features, and then performs a turning action at that position. It walks around the boundary determined by the visual reference data. During the process of walking around the boundary determined by the visual reference data, the recording device is triggered to record the coordinates of the current position obtained by the laser SLAM at a preset time interval or distance interval. This coordinate point is marked as a boundary point and recorded in the storage unit of the device.

[0063] Step 3: After the robot completes one cycle along the boundary determined by the visual reference data or returns to the base station position, mapping is completed. At this point, the coordinates of the robot's working boundary will be fully recorded. The coordinates of each boundary point can then be fitted or directly connected sequentially to form a map. Figure 9 The closed operation map shown. Figure 9 The white area represents the map area detected by the LiDAR, with the central line indicating the boundary of the work area. The coordinates of the upper boundary of the work map are all coordinates from the LiDAR SLAM mapping. After the map is completed, the robot can be configured to traverse and perform tasks within the defined work area on the map.

[0064] In step 2, after leaving the station, in addition to walking randomly to a position where the visual reference data matches the boundary features, the above-mentioned equipment can be further configured to: after the laser SLAM function is turned on, manually remotely control the machine to walk around the grass boundary, and during this period, record the boundary coordinates determined by the laser reference data at the current position according to the preset time interval or distance interval, and create an operation map containing only grass, so that in the subsequent operation process, the machine can plan a path in the grass according to the boundary range defined by the map under the positioning of the laser radar, and traverse and mow the grass.

[0065] Furthermore, when the outdoor robot's base station is located on the boundary of the work area, the device can be configured to, after leaving the station in step 2, first retreat to the left of the base station and then turn to the right, or the outdoor robot can be configured to first walk in a straight line in a single direction after leaving the station, actively walking to a position where the visual reference data on the side of the base station matches the boundary features. Then, according to the walking direction of the visual SLAM calibration device, it is made to walk around the boundary line once, while the coordinates of the machine under laser SLAM are recorded at 10cm intervals.

[0066] Therefore, once the machine has completed a full circuit along the grassland boundary and returned to the charging station, the coordinates of its working grassland boundary can be accurately recorded. This creates a closed operational map. During subsequent operation, the robot plans its work path within this closed map, and laser positioning controls its trajectory. Strictly adhering to the planned path improves the machine's operational efficiency within its work area.

[0067] In summary, this application, based on LiDAR and machine vision, can efficiently and accurately acquire the boundary of the working area of ​​an outdoor self-moving robot through the collaboration between LiDAR SLAM and visual SLAM, thereby improving the accuracy of the machine's path planning function and enhancing its lawn mowing performance.

[0068] This application achieves accurate determination of the boundary area through visual SLAM, thereby triggering laser SLAM to provide precise coordinates of the boundary point position. This enables the application to overcome the positioning deviation of visual SLAM, avoid laser SLAM from mistakenly including the flat road surface outside the working area into the boundary area due to its inability to identify the specific ground conditions, and effectively avoid the deviation of sensor signals, delay or interference of satellite signals under other external positioning methods.

[0069] This application allows for flexible selection of the specific method for finding and locating boundary coordinates during the mapping process, based on the characteristics of the work area and the operational requirements of the equipment. It avoids blind spots in path planning through random traversal, improves mapping efficiency by running along boundary lines, and prevents the machine from straying out of the work area at specific walking angles by directly utilizing image information acquired through visual SLAM and setting the machine's turning method.

[0070] Compared to current mainstream navigation methods such as GPS, laser SLAM, and VSLAM, GPS navigation offers the highest accuracy but has high equipment requirements and costs. It requires a network or satellite signal and its accuracy is poor in areas with weak communication. Therefore, it requires an open environment. In urban environments, trees and tall buildings can affect GPS accuracy. Laser SLAM and visual VSLAM have lower equipment requirements and costs, and they do not have network communication issues, allowing the device to operate even without an internet connection. They also offer relatively high accuracy in urban environments. However, the combination of the aforementioned mainstream navigation methods has the following drawbacks compared to the solution described in this application:

[0071] The combination of laser SLAM and GPS: Although it has the highest positioning accuracy and can guarantee good positioning accuracy under various conditions, it does not have the ability to identify grassland boundaries. The boundary lines need to be set manually and cannot be automatically obtained.

[0072] The combination of visual SLAM and GPS: Although GPS positioning is highly accurate and boundary recognition can be achieved through visual SLAM, in urban environments, GPS signals are easily lost or weak, which can lead to a significant drop in positioning accuracy. At the same time, visual SLAM has poor positioning accuracy in non-boundary areas (such as completely grassy areas). Therefore, in practical applications, it often fails to accurately locate and causes navigation to deviate.

[0073] The laser SLAM + visual SLAM adopted in this application provides high positioning accuracy using laser SLAM, essentially guaranteeing accurate positioning at all times in urban conditions, and also achieving lower cost compared to GPS. When the equipment reaches the boundary of the work area, the visual SLAM can actively trigger the laser SLAM for precise positioning and excite the boundary point coordinates after detecting boundary features. Therefore, this application can automatically generate a more accurate work map, ensuring reliable subsequent operation of the equipment, preventing it from operating outside the boundary, and protecting the safety of equipment, personnel, and property.

[0074] The above are merely embodiments of this application, and their descriptions are quite specific and detailed, but they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application.

Claims

1. A method for working area delineation based on a laser and vision scheme, characterized by the steps of Comprising: In the process of walking, the visual reference data is obtained by the visual sensing unit, and the laser reference data is obtained by the laser sensing unit; When the visual reference data meets the boundary feature, the boundary coordinates determined according to the laser reference data at the current position are recorded; When the mapping requirement is met, the working area of the outdoor robot is circled according to the boundary coordinates; In the process of walking, the following steps are performed to execute the turning action: According to the difference between the proportions of the left and right non-grass areas in the environment picture collected by the visual sensing unit, if the difference exceeds the preset proportion, turn to the side with smaller proportion of non-grass area; if the difference does not exceed the preset standard, turn according to the preset rule; The preset rule includes any of the following turning modes: Turn according to the preset fixed direction, or turn randomly.

2. The method of claim 1, wherein the method is a laser and vision based approach for working area delineation, characterized in that, The mapping requirement meets any of the following conditions or their combination: In the process of walking, the map grid corresponding to the boundary coordinates no longer increases; The boundary coordinates can be fitted to form a closed figure; The walking time of the outdoor robot to determine the boundary coordinates reaches the preset time; 3. The method of claim 1, wherein the method is a laser and vision based work area delineation method, further comprising: The outdoor robot walks along the boundary determined by the visual reference data for one round.

4. The method of claim 1, wherein the method is a laser and vision based work area delineation method, further comprising: In the process of walking, the walking direction is randomly selected, and the turning action is executed when the visual reference data meets the boundary feature.

5. The method of claim 4, wherein the laser and vision based work area delineation method further comprises: In the process of walking, the outdoor robot first walks to the position where the visual reference data meets the boundary feature, and then executes the turning action at this position and walks along the boundary determined by the visual reference data for one round.

6. The method of claim 4, wherein the laser and vision based work area delineation method further comprises: In the process of walking along the boundary determined by the visual reference data for one round, the boundary coordinates determined according to the laser reference data at the current position are recorded at preset time intervals or distance intervals. In the process of walking, the outdoor robot first walks to the position where the visual reference data meets the boundary feature in any of the following ways: when the base station of the outdoor robot is located on the boundary of the working area, the outdoor robot is set to walk back and turn after leaving the station, and then walk to the position where the visual reference data meets the boundary feature on the side of the base station; 7. The method of claim 1, wherein the method is a laser and vision based work area delineation method, further comprising: The outdoor robot is set to walk in a straight line in a selected direction after leaving the station, and then walk to the position where the visual reference data meets the boundary feature.

8. An outdoor robot, characterized in that, The turning angle is set to 90°-270°. Comprising: A visual sensing unit for obtaining visual reference data; A laser sensing unit for obtaining laser reference data; A first storage unit for storing computer programs or instructions; 9. The outdoor robot of claim 8, wherein, A control unit for executing the computer programs or instructions in the storage unit, so that the outdoor robot executes the working area circumscription method based on the laser and visual scheme according to any of claims 1-7. Further comprising a second storage unit for storing the boundary of the working area of the outdoor robot; In the process of outdoor robot operation, the control unit is further used to compare whether the position coordinates determined by the laser reference data reach the boundary of the working area and / or whether the visual reference data meets the boundary feature, and when the position coordinates reach the boundary of the working area and / or the visual reference data meets the boundary feature, the turning action of the outdoor robot is triggered.

Images