Mowing robot and control method thereof

Abstract

The invention discloses a mowing robot and a control method thereof, and relates to the technical field of robots, and the robot comprises a driving module, an operation module, a navigation module, an energy module and a frame; the navigation module comprises two RTK antennas, a laser radar and a calculation module; the calculation module is used for determining the position and direction of the mowing robot as first positioning information by using satellite signals received by the two RTK antennas and correction data sent by the base station; when the error of the first positioning information is greater than a first set value, constructing a voxel map through a point cloud obtained through laser radar scanning, and determining the position and direction of the mowing robot as second positioning information according to the voxel map; planning according to the second positioning information to obtain a global path; the driving module is controlled to drive the mowing robot to move according to the global path; and controlling the operation module to work for mowing. According to the invention, the positioning precision is improved through the RTK antenna or the laser radar; and the calculation module can adapt to changeable terrains.

Description

Technical Field

[0001] This application relates to the field of robot technology, and particularly to a lawn mowing robot and its control method. Background Art

[0002] With the acceleration of the urbanization process and the expansion of the greening area, the traditional manual lawn mowing method can no longer meet the needs of rapid development. Especially in large lawns, gardens, parks, sports fields and other places, the traditional lawn mowing means have problems such as low efficiency, high labor cost and unsafe working environment.

[0003] Currently, some existing intelligent lawn mowing robots have most of their application scenarios limited to small lawns or home gardens. For the mowing needs of large lawns and large areas of green land, the existing technologies still have problems such as low accuracy of the navigation system and limited ability to adapt to complex terrains. Summary of the Invention

[0004] The main purpose of the embodiments of this application is to propose a lawn mowing robot and its control method to improve the navigation accuracy of the lawn mowing robot and thus improve its ability to adapt to complex terrains.

[0005] To achieve the above object, on the one hand, an embodiment of this application proposes a lawn mowing robot, which includes: a driving module, an operation module, a navigation module, an energy module and a vehicle frame; Wherein, the driving module, the operation module, the navigation module and the energy module are all installed on the vehicle frame; The navigation module includes two RTK antennas, a lidar and a calculation module; The calculation module includes: A first positioning unit, configured to determine the position and direction of the lawn mowing robot as first positioning information by using the satellite signals received by the two RTK antennas and the correction data sent by the reference station; A second positioning unit, configured to, when the error of the first positioning information is greater than a first set value, construct a voxel map through the point cloud scanned by the lidar, and then determine the position and direction of the lawn mowing robot as second positioning information according to the voxel map; A path planning unit, configured to plan a global path according to the first positioning information or the second positioning information; A driving control unit, configured to control the driving module to drive the lawn mowing robot to move along the global path; An operation control unit, configured to control the operation module to work for lawn mowing.

[0006] In some embodiments, the second positioning unit includes: A preprocessing unit for determining undistorted point clouds using IMU data measured by the IMU unit in the navigation module; A state estimation unit for optimizing using a factor graph and registering the undistorted point clouds as priors to the point clouds; and then dynamically registering the scanned point clouds to the voxel map; A point cloud matching and optimization unit for updating the matching between the point clouds and the voxel map using an extended Kalman filter method to optimize the pose of the lawn mowing robot; calculating the residuals from the point clouds to the voxel map, and minimizing the error from the point clouds to the voxel map under the constraint of the IMU unit; A local map update unit for storing the voxel map closest to the current time and within a set time period in a sliding window manner; A loop update unit for identifying whether the lawn mowing robot has been to the same location to eliminate cumulative errors, optimize pose estimation, and reposition.

[0007] In some embodiments, the loop update unit includes: A key frame management unit for storing the point cloud data and pose at the corresponding time as key frames after a set period of time has elapsed or when the pose change of the lawn mowing robot exceeds a preset threshold; A frame search unit for searching for candidate key frames similar to the current key frame in the stored key frames using a KD-Tree; A point cloud matching unit for calculating the ICP matching between the candidate key frame and the current key frame to obtain a relative transformation matrix and an ICP error; A factor graph optimization unit for adding a loop constraint to the factor graph if the ICP error is less than a second set value; using iSAM2 incremental optimization for global optimization to correct the drift error of the voxel map.

[0008] In some embodiments, the path planning unit includes: A boundary generation unit for generating a working boundary based on the first positioning information or the second positioning information; A path generation unit for generating parallel lines at intervals of the radius of the cutter head in the working boundary, and then connecting the heads and tails of the parallel lines to form a continuous global path to completely cover the working boundary.

[0009] In some embodiments, the drive control unit includes: A look-ahead point selection unit for dynamically selecting look-ahead points on the global path such that the look-ahead points maintain a set distance from the current position of the lawn mowing robot until the look-ahead points coincide with the end point; A direction adjustment unit for dynamically adjusting the direction of the mowing robot and controlling the drive module to drive the mowing robot to move towards the look-ahead point.

[0010] In some embodiments, the drive control unit further includes an obstacle avoidance unit, and the obstacle avoidance unit includes: An obstacle recognition unit for determining the object surface according to the return time of the lidar beam or the angle between the incident beam and the return beam, projecting the object surface onto the global map to identify whether the object is an obstacle; or, using the camera module in the navigation module to obtain an image, segmenting the image through an AI model, projecting the segmented object onto the global map, and then determining whether the object is an obstacle; A temporary path generation unit for generating a temporary path when an obstacle is recognized; A bypass unit for, when an obstacle is recognized, expanding the boundary of the obstacle by a set range, and then controlling the mowing robot to move according to the temporary path to bypass the obstacle after the boundary expansion, and then return to the global path.

[0011] In some embodiments, the temporary path generation unit includes: A feasible speed space generation unit for calculating and generating the feasible speed space of the mowing robot within a dynamic time window according to the kinematic constraints of the mowing robot; A trajectory prediction unit for predicting the predicted trajectory corresponding to each speed combination within a preset time period in the feasible speed space; wherein, the speed combination includes a linear velocity and an angular velocity; A trajectory scoring unit for scoring the corresponding predicted trajectory according to the distance between the end point of the predicted trajectory and the target point, the forward speed of the predicted trajectory, the length of the predicted trajectory, and the distance between the predicted trajectory and the obstacle; A trajectory selection unit for determining the predicted trajectory with the highest score and then generating the temporary path corresponding to the speed combination.

[0012] In some embodiments, the obstacle avoidance unit further includes: An obstacle boundary judgment unit for judging whether the space between the obstacle and the working boundary in the global map is greater than the width of the mowing robot to judge whether bypassing the obstacle will exceed the working boundary; if bypassing the obstacle will exceed the working boundary, interrupt the current path, return to the starting point of the current path, and then replan the navigation path of the remaining unworked area after subtracting the worked area.

[0013] In some embodiments, the energy module includes multiple batteries, and the battery compartments of each battery adopt a quick-release structure; The calculation module further includes: An energy management unit, configured to switch the currently powered battery to the remaining batteries if the current of the currently powered battery is lower than a preset power threshold, and send a message reminding to replace the battery to the management terminal of the lawn mowing robot.

[0014] To achieve the above object, on the other hand, an embodiment of the present application provides a control method, which includes the following steps: Determine the position and orientation of the lawn mowing robot as the first positioning information by using the satellite signals received by two RTK antennas and the correction data sent by the reference station; When the error of the first positioning information is greater than a first set value, construct a voxel map through the point cloud obtained by lidar scanning, and then determine the position and orientation of the lawn mowing robot as the second positioning information according to the voxel map; Plan a global path according to the first positioning information or the second positioning information; Control the driving module to drive the lawn mowing robot to move along the global path; Control the operation module to work for mowing the lawn.

[0015] The embodiment of the present application has at least the following beneficial effects: The lawn mowing robot of the present application includes: a driving module, an operation module, a navigation module, an energy module and a frame; the navigation module includes two RTK antennas, a lidar and a calculation module; the calculation module includes: a first positioning unit, configured to determine the position and orientation of the lawn mowing robot as the first positioning information by using the satellite signals received by two RTK antennas and the correction data sent by the reference station; a second positioning unit, configured to construct a voxel map through the point cloud obtained by lidar scanning when the error of the first positioning information is greater than a first set value, and then determine the position and orientation of the lawn mowing robot as the second positioning information according to the voxel map; a path planning unit, configured to plan a global path according to the first positioning information or the second positioning information; a driving control unit, configured to control the driving module to drive the lawn mowing robot to move along the global path; an operation control unit, configured to control the operation module to work for mowing the lawn. The present application performs positioning through RTK antennas or lidar, improves the positioning accuracy, and further provides accurate position information for navigation; by planning the global path through the calculation module, it can adapt to changing terrains and improve the adaptability to complex terrains. Description of the Drawings

[0016] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on these drawings.

[0017] Figure 1 An exemplary structural diagram of a lawn mowing robot provided by an embodiment of the present application; Figure 2 A schematic flowchart of a control method provided by an embodiment of the present application; Figure 3(a) is a top view of a lawn mowing robot provided by an embodiment of the present application; Figure 3(b) is an axonometric view of a lawn mowing robot provided by an embodiment of the present application; Figure 3(c) is a front view of a lawn mowing robot provided by an embodiment of the present application; Figure 3(d) is a side view of a lawn mowing robot provided by an embodiment of the present application.

[0018] Reference numerals: 101 is a cutter, 102 is a lifting chassis, 103 is a crawler, 104 is an RTK antenna hole position, 105 is a battery compartment, 106 is a video transmission camera, 107 is a lidar, and 108 is a drive wheel. Detailed implementation manners

[0019] In order to make the objectives, technical solutions and advantages of the present application more clear and understandable, the following further details the present application in combination with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application. When the following description involves the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementation manners described in the following exemplary embodiments do not represent all implementation manners consistent with the embodiments of the present application. They are only examples of devices and methods consistent with some aspects of the embodiments of the present application detailed in the appended claims.

[0020] It can be understood that the terms "first", "second", etc. used in the present application can be used in this document to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present application, the first information can also be referred to as the second information, and similarly, the second information can also be referred to as the first information. Depending on the context, the words "if", "when" as used herein can be interpreted as "when...", "when...", or "in response to a determination".

[0021] As used in this application, terms such as "at least one", "a plurality", "each", "any one", etc. The "at least one" includes one, two or more than two, the "a plurality" includes two or more than two, "each" refers to each one in the corresponding plurality, and "any one" refers to any one in the plurality.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0023] Before elaborating on the embodiments of this application in detail, some related technologies involved in the embodiments of this application are described as follows: RTK: Real - time kinematic, which is a real - time kinematic carrier phase differential technology.

[0024] LiDAR (Light Laser Detection and Ranging), which is short for a laser detection and ranging system.

[0025] SLAM: Simultaneous Localization and Mapping (abbreviated as SLAM).

[0026] Loop detection, also known as closed - loop detection, refers to the ability of a robot to recognize that it has reached a certain scene, enabling the map to form a closed loop.

[0027] PID control algorithm: PID is the abbreviation of Proportional, Integral, and Differential. The PID control algorithm is a control algorithm that combines the three links of proportional, integral, and differential.

[0028] Voxel map: A voxel (Volume Pixel) is a volume pixel in three - dimensional space, similar to a pixel in a two - dimensional image but extended to three dimensions. A voxel map divides the environment into a uniform cubic grid, and each cube (voxel) stores information within that space (such as occupancy probability, color, semantic label, etc.).

[0029] ESKF, namely Extended Kalman Filter: An improved Kalman filter algorithm that focuses on estimating the error of the state (such as pose error) rather than directly estimating the system state (such as position and attitude). By linearizing the error dynamic model, it solves the state estimation problem of non - linear systems.

[0030] KD - Tree: A binary tree data structure used to efficiently organize points in k - dimensional space. By recursively partitioning the space (selecting median points in different dimensions), it accelerates the Nearest Neighbor Search.

[0031] ICP Matching: An iterative algorithm used to align two point clouds (or a point cloud and a map), minimizing the distance error between them. The core steps include: 1. Association: Finding the nearest neighbor points in the target point cloud.

[0032] 2. Solving the transformation: Calculating the rotation and translation matrices through SVD decomposition.

[0033] 3. Iterative optimization: Repeating until convergence.

[0034] iSAM2: An incremental optimization framework based on Factor Graph for SLAM backend optimization. It efficiently updates pose and map variables through Bayes Tree.

[0035] Robot dynamics model: A mathematical model that describes the relationship between the motion of a robot and forces / torques, usually based on Newton - Euler equations or Lagrangian mechanics.

[0036] In view of the problems existing in the prior art, some embodiments of the present application aim to solve the following technical problems: For large - scale special operation robots, the need for all - weather and all - terrain traveling and operating capabilities will inevitably bring problems of the volume and weight of the overall structure. An overly long body and ground contact length will bring excellent passability, but their load - bearing and friction during steering will cause a certain degree of damage and compaction to the soil. Subsequently, this situation can be alleviated by adjusting the motor output and torque, and reducing the machine weight can also achieve this.

[0037] Based on this, the solution of the present application proposes a lawn mowing robot and its control method, aiming to achieve efficient and precise lawn mowing operations for large - area lawns through advanced intelligent control technology, an efficient battery system, and multi - sensor collaborative work, with strong adaptability, long battery life, and environmental friendliness.

[0038] Refer to Figure 1 , the embodiments of the present application provide a lawn mowing robot, which includes: a driving module, an operation module, a navigation module, an energy module, and a frame; Among them, the driving module, the operation module, the navigation module, and the energy module are all installed on the frame; The navigation module includes two RTK antennas, a lidar, and a calculation module; The calculation module includes: The first positioning unit is used to determine the position and orientation of the lawn mowing robot as the first positioning information by using the satellite signals received by the two RTK antennas and the correction data sent by the reference station; The second positioning unit is used to, when the error of the first positioning information is greater than a first set value, construct a voxel map through the point cloud obtained by lidar scanning, and then determine the position and orientation of the lawn mowing robot as the second positioning information according to the voxel map; The path planning unit is used to plan a global path according to the first positioning information or the second positioning information; The drive control unit is used to control the drive module to drive the lawn mowing robot to move along the global path; The operation control unit is used to control the operation module to work for lawn mowing.

[0039] As an optional implementation manner, the second positioning unit includes: The preprocessing unit is used to determine the undistorted point cloud by using the IMU data measured by the IMU unit in the navigation module; The state estimation unit is used to optimize by using a factor graph and register the undistorted point cloud as a prior to the point cloud; and then dynamically register the scanned point cloud to the voxel map; The point cloud matching and optimization unit is used to update the matching between the point cloud and the voxel map by using the extended Kalman filter method to optimize the pose of the lawn mowing robot; calculate the residual from the point cloud to the voxel map, and minimize the error from the point cloud to the voxel map under the constraint of the IMU unit; The local map update unit is used to store the voxel map closest to the current moment and within a set time period in a sliding window manner; The loop update unit is used to identify whether the lawn mowing robot has ever reached the same location to eliminate cumulative errors, optimize pose estimation and repositioning.

[0040] Furthermore, the loop update unit includes: The key frame management unit is used to store the point cloud data and pose at the corresponding moment as key frames after a set period of time or when the pose change of the lawn mowing robot exceeds a preset threshold; The frame search unit uses a KD-Tree to search for candidate key frames similar to the current key frame in the stored key frames; The point cloud matching unit is used to calculate the ICP matching between the candidate key frame and the current key frame to obtain a relative transformation matrix and an ICP error; A factor graph optimization unit, which is used to add loop constraints to the factor graph if the ICP error is less than a second set value; and perform global optimization using iSAM2 incremental optimization to correct the drift error of the voxel map.

[0041] As an alternative implementation, the path planning unit includes: A boundary generation unit, which is used to generate a working boundary according to the first positioning information or the second positioning information; A path generation unit, which is used to generate parallel lines at intervals of the radius of the cutter head in the working module within the working boundary, and then connect the heads and tails of each parallel line to form the continuous global path to completely cover the working boundary.

[0042] As an alternative implementation, the drive control unit includes: A look-ahead point selection unit, which is used to dynamically select look-ahead points on the global path so that the look-ahead points maintain a set distance from the current position of the mowing robot until the look-ahead points coincide with the end point; A direction adjustment unit, which is used to dynamically adjust the direction of the mowing robot and control the drive module to drive the mowing robot to move towards the look-ahead point.

[0043] As an alternative implementation, the drive control unit further includes an obstacle avoidance unit, and the obstacle avoidance unit includes: An obstacle recognition unit, which is used to determine the object surface according to the return time of the lidar beam or the angle between the incident beam and the return beam, project the object surface onto the global map and then identify whether the object is an obstacle; or, use the camera module in the navigation module to obtain an image, segment the image through an AI model, project the segmented object onto the global map, and then determine whether the object is an obstacle; A temporary path generation unit, which is used to generate a temporary path when an obstacle is recognized; A detour unit, which is used to expand the boundary of the obstacle by a set range when an obstacle is recognized, and then control the mowing robot to move according to the temporary path to bypass the obstacle after the boundary expansion, and then return to the global path.

[0044] Furthermore, the temporary path generation unit includes: A feasible speed space generation unit, which is used to calculate and generate the feasible speed space of the mowing robot within a dynamic time window according to the kinematic constraints of the mowing robot; A trajectory prediction unit, which is used to predict the prediction trajectory corresponding to each speed combination within a preset time period in the feasible speed space; wherein, the speed combination includes a linear speed and an angular speed; A trajectory scoring unit, configured to score a corresponding predicted trajectory according to the distance between the end point of the predicted trajectory and a target point, the forward speed of the predicted trajectory, the length of the predicted trajectory, and the distance between the predicted trajectory and an obstacle; A trajectory selection unit, configured to determine the predicted trajectory with the highest score and then generate the temporary path from the corresponding speed combination.

[0045] As another further embodiment, the obstacle avoidance unit further includes: An obstacle boundary judgment unit, configured to judge whether the space between the obstacle and the working boundary in the global map is greater than the width of the lawn mowing robot, so as to judge whether bypassing the obstacle will exceed the working boundary; if bypassing the obstacle will exceed the working boundary, interrupt the current path, return to the starting point of the current path, and then re-plan the navigation path of the remaining unworked area after subtracting the worked area.

[0046] As an optional embodiment, the energy module includes multiple batteries, and the battery compartments of each battery adopt a quick-release structure; The computing module further includes: An energy management unit, configured to switch the currently powered battery to the other batteries if the current of the currently powered battery is lower than a preset power threshold, and send a message reminding to replace the battery to the management terminal of the lawn mowing robot.

[0047] Refer to Figure 2 , this application also provides a control method, and the control method includes the following steps S1 to S5: S1: Using the satellite signals received by two RTK antennas and the correction data sent by a reference station to determine the position and orientation of the lawn mowing robot as the first positioning information; S2: When the error of the first positioning information is greater than a first set value, construct a voxel map from the point cloud obtained by lidar scanning, and then determine the position and orientation of the lawn mowing robot according to the voxel map as the second positioning information; S3: Plan a global path according to the first positioning information or the second positioning information; S4: Control the driving module to drive the lawn mowing robot to move along the global path; S5: Control the operation module to work for lawn mowing.

[0048] Next, the solution of the embodiment of this application will be introduced and described in detail with specific application examples.

[0049] Referring to FIGS. 3(a), 3(b), 3(c), and 3(d), where FIG. 3(a) is a top view of the lawn mowing robot, FIG. 3(b) is a perspective view of the lawn mowing robot, FIG. 3(c) is a front view of the lawn mowing robot, and FIG. 3(d) is a side view of the lawn mowing robot. The lawn mowing robot of this embodiment may include a drive module, an operation module, a navigation module, and an energy module, and the above four modules are all built on the vehicle frame. The drive module uses two vertical motors connected directly to two horizontal reducers, and the drive wheels drive the crawlers located on both sides of the machine. The operation module is located at the bottom of the vehicle frame and is arranged asymmetrically with two motors. The electric push rod lifting platform can move with a clearance of 2.5 cm - 13.5 cm from the ground. The navigation module is divided into two parts, a sensor and a calculation module; the sensors are located at the top and the head of the vehicle shell, which are an RTK antenna, a lidar, and a camera module respectively. The RTK antenna receives satellite signals for positioning and navigation. The lidar strengthens the positioning when approaching the woods or when there are other environmental obstructions, and at the same time works together with the camera module for obstacle recognition and avoidance. The calculation module is located under the vehicle shell and is placed on the vehicle frame, processing satellite signals and obstacle recognition, and running navigation locally. The energy module is located at the front end of the vehicle shell and is blocked by a hatch cover. It is two lithium iron phosphate batteries, which can be automatically switched when one of them has insufficient power, and a single person can perform disassembly and battery replacement operations.

[0050] I. Manual Remote Control: Connect the robot through the remote control. The left joystick can be used to control the movement and steering of the robot, and the roller on the upper right side of the remote control can be toggled to manually adjust the height of the cutter head.

[0051] II. Automatic Operation: 1) Toggle the right lever to turn on the automatic operation mode.

[0052] 2) Connect and bind the robot, and wait until the robot automatically receives the RTK signal and calibrates its real-time position by combining with the measurement station analysis.

[0053] 3) Click on the device details and select the path planning method (zigzag or bow-shaped).

[0054] 4) Move the robot and click on the new point to mark points at multiple corners of the operation area.

[0055] 5) If there is a no-go area within the area, after clicking on the no-go area, repeat the operation in 4) to add points for the no-go area.

[0056] 6) After completing the marking of the working area and the no-go area, click on the return point. After moving the robot to the specified position, click on the next step to determine the starting point and the return point.

[0057] 7) Click on "Complete" to complete the map drawing.

[0058] 8) Click Start Job to start the automation job on the selected map.

[0059] Specifically, this embodiment provides an intelligent electric lawn mowing robot for large-area lawn maintenance, which integrates high-precision RTK positioning technology to achieve autonomous navigation, obstacle avoidance and precise mowing.

[0060] 1. Robot startup and environmental perception: 1. The RTK antenna receives the carrier phase signal of satellite systems such as GPS / Beidou. The base station (with known precise coordinates) compares the received satellite signal with its own position, generates differential correction data, and transmits it to the mobile station (robot) in real time via radio or network. The mobile station combines the satellite signal it receives with the correction data of the base station to eliminate errors such as atmospheric delay and satellite clock error, solve the integer ambiguity of the carrier phase, and finally achieve centimeter-level positioning (±2cm). Due to the use of dual antennas, the position and direction of the robot can be calculated based on the installation position of the two antennas and the positioning information of the two antennas. At the same time, the robot continuously transmits the positioning data (latitude, longitude, elevation) to the navigation module in real time for path planning adjustment and position calibration.

[0061] 2. When there are many buildings or trees around the robot, resulting in poor RTK data quality, laser simultaneous localization and mapping (SLAM) technology can be used to provide mapping and positioning information.

[0062] 1) Preprocessing stage: IMU pre-integration: Use IMU data to predict the initial pose of the current frame.

[0063] Point cloud dedistortion: The point cloud distortion generated during the scanning process is compensated according to the IMU motion, so that the point cloud returns to the unified time coordinate.

[0064] 2) Recursive state estimation: Factor graph optimization is used to efficiently register point clouds using IMU odometry predictions as priors.

[0065] Directly register new scan points to the voxel map to avoid building a complete point cloud frame, thereby reducing the amount of calculation.

[0066] 3) Point cloud matching and optimization: Compared with frame-to-frame matching, this embodiment can directly match the laser point with the point in the local map: ESKF (Extended Kalman Filter) is used to update the state and optimize the posture.

[0067] Calculate the residual from the laser point to the map and minimize the error under the IMU constraints.

[0068] 4) Recursive local map update: Use a voxel map, only keep the key point cloud instead of the complete point cloud.

[0069] Manage the local map in a sliding window manner, only store the point cloud in the recent period of time, prevent the map from expanding, and reduce the storage requirements of the lawn mowing robot.

[0070] 5) Loop closure update: The above process can only ensure the accuracy of the local map. When the site is large, the error accumulation of the local map will lead to a large final error of the global map. To solve the above problems, this embodiment can perform loop closure detection and loop closure update. Through loop closure update, the cumulative error generated by local map update can be eliminated, pose estimation and relocalization can be optimized. The steps of loop closure detection are as follows: a. Key frame management: Only store key frames (point cloud data and poses at critical moments), rather than all point cloud frames, to reduce the computational amount.

[0071] Key frames are usually added at regular intervals or when there is a large enough pose change.

[0072] b. Approximate nearest neighbor search (ANN): Use a KD-Tree to search for similar point cloud frames in the historical key frame database to find possible candidate key frames.

[0073] c. Point cloud matching (ICP): Calculate the ICP matching between the candidate key frame and the current key frame to obtain the relative transformation matrix (discard it when the error is large).

[0074] d. Factor graph optimization: If the ICP error is small, add the loop closure constraint to the factor graph (the pose constraint between loop closure frames).

[0075] Use iSAM2 incremental optimization for global optimization to correct the drift error of the entire map.

[0076] Third, while gradually constructing the map, analyze the single-frame point cloud, transform the point cloud into the robot coordinate system. When the height of the point cloud is higher than the set lawn height, it is considered an obstacle. Then map the detected obstacle position to the global map and mark it as an avoidance area.

[0077] 2. Autonomous path planning and mowing operation: 1. Receive the positioning information provided by RTK or laser SLAM, generate a working boundary, generate parallel lines at intervals of the cutter head radius within the working boundary, connect the heads and tails of these parallel lines to form a continuous path (global path) that completely covers the polygon surrounded by the working boundary, so as to ensure that the cutter covers all areas inside the boundary during the robot's walking process. Then, on the pre-planned trajectory, select a target point within a certain range from the current vehicle, called the look-ahead point, continuously adjust the direction of the robot so that it walks towards the look-ahead point. The look-ahead point continuously slides on the predetermined trajectory and maintains a certain distance from the robot until it coincides with the end point, then the robot can walk along the path.

[0078] 2. During the mowing operation, the lidar emits a laser beam, which returns after reaching the object surface. According to the time of laser return or the angle between the incident and return beams, the position of the object surface (a point in space) is determined. The radar simultaneously emits a series of beams, generating a series of point clouds in space, and these point clouds correspond to the object surface contour. The camera module can also identify obstacles, which are generally relatively low (the radar cannot distinguish the ground from low obstacles), and through the AI model for image segmentation, after segmenting the obstacles, they will be projected onto the ground. When the robot detects an object blocking the walking path, it will generate a temporary path to avoid the blocking object. This path ensures that there is no collision between the robot contour and the object point cloud during the walking process and is as close as possible to the global path, and returns to the global path as soon as possible after the obstacle avoidance is completed.

[0079] In this embodiment, by generating a temporary path, the mowing robot can bypass obstacles when moving along the global path and return to the global path as soon as possible after the obstacle avoidance is completed, avoiding re-planning a new global path due to detouring and not moving along the global path.

[0080] The generation steps of the temporary path include: 1) Generate a feasible velocity space: The robot can choose different combinations of linear velocity and angular velocity at each moment.

[0081] Calculate the feasible velocity space: Robot kinematic constraints (maximum acceleration, maximum velocity, maximum angular velocity, maximum angular acceleration).

[0082] The velocity range that the robot can reach within the dynamic window (within a limited time).

[0083] 2) Predict the trajectory: In the feasible velocity space, predict the trajectory corresponding to each velocity combination.

[0084] The time for trajectory simulation is usually 1 - 2 seconds to evaluate the motion effect in the short term.

[0085] 3) Evaluate the trajectory score: Calculate the comprehensive score for each trajectory, mainly considering: Heading: The distance between the end point of the trajectory and the target point (the closer to the target, the higher the score. This target point is a point on the path that has not been executed yet and is not occupied by obstacles. If all the paths that have not been executed are occupied, then the next path is selected in turn).

[0086] Velocity: The forward speed corresponding to the trajectory (the faster the speed, the higher the score).

[0087] Path: The shorter the detour path, the higher the score.

[0088] Clearance: The distance between the trajectory and the obstacle (the safer the trajectory, the higher the score).

[0089] The final score is obtained through weighted calculation.

[0090] 4) Select the optimal trajectory: Select the trajectory with the highest score as the motion command (linear velocity and angular velocity) at the current moment.

[0091] The robot executes this command and enters the next loop.

[0092] III. Through the real-time pose analysis of the machine, adjust the raising and lowering of the cutter head lifting motor to ensure as much as possible to fit the ground and perform mowing operations at the set height. When the lidar and camera module detect and identify obstacles, while avoiding them, reduce the rotation speed of the cutter to avoid uneven mowing effects caused by repeated operations on the areas where mowing has been completed.

[0093] 3. Obstacle avoidance and dynamic adjustment: I. During mowing operations, the lidar emits a laser beam, which returns after reaching the surface of an object. According to the time of laser return or the angle between the incident and return beams, the position of the object surface (a point in space) is determined. The radar also emits a series of beams, generating a series of point clouds in space, and these point clouds will correspond to the object surface contour.

[0094] II. Project the obstacle information identified and segmented by the lidar and camera module onto the global map, and at the same time expand the boundary of the obstacle by 10 - 15 cm to ensure that the robot body can maintain a safe distance from the obstacle (10 cm for stationary obstacles and 15 cm for moving obstacles) at all times. At the same time, judge whether the space of the obstacle from the boundary in the global map is greater than the width of the machine, and whether detouring will exceed the working boundary. If the obstacle has enough distance from the boundary to allow the machine to detour, the navigation algorithm will plan a detour route as close to the edge as possible based on the boundary of the obstacle, avoiding missed cutting while avoiding obstacles. If the distance between the obstacle and the working boundary is less than the width of the machine and detouring is not supported, the current path will be interrupted, and the robot will return to the starting point of the current path. After subtracting the worked area, the navigation path for the remaining unworked area will be replanned.

[0095] 4. Battery life management and recharging: I. When the battery level of a single battery is lower than 5%, the robot automatically switches to the backup battery, and at the same time the APP reminds the staff to perform battery replacement operations.

[0096] II. It can supply power to multiple batteries simultaneously. The battery compartment of the robot adopts a quick-release structure, which facilitates battery replacement operations and enables longer working hours.

[0097] The beneficial effects of this embodiment include: Adopting high-precision RTK positioning technology can achieve precise positioning even in very open environments (such as dams).

[0098] Intelligent path planning is realized, avoiding repeated coverage and omitted areas, and improving work efficiency.

[0099] Through multi-sensor fusion for obstacle avoidance, safety is improved, and it can work stably in complex environments.

[0100] The mowing height can be adjusted to meet the needs of different types of lawns and improve the mowing quality.

[0101] Remote management and data analysis can be used to monitor the working status in real time through the APP and perform intelligent scheduling.

[0102] The embodiments described in this application are to more clearly illustrate the technical solutions of this application, and do not constitute a limitation to the technical solutions provided by this application. Those skilled in the art know that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by this application are equally applicable to similar technical problems.

[0103] Those skilled in the art can understand that the technical solutions shown in the figures do not constitute a limitation to this application. It may include more or fewer technical solutions than those shown in the figures, or combine some technical solutions, or different technical solutions.

[0104] In the description of this application and the above-mentioned accompanying drawings, terms such as "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of the present application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0105] It should be understood that in this application, "at least one (item)" means one or more, and "a plurality" means two or more. "And / or" is used to describe the association relationship of associated objects and indicates that three relationships can exist. For example, "A and / or B" can mean: only A exists, only B exists, and both A and B exist at the same time. Among them, A and B can be singular or plural. The character " / " generally means that the associated objects before and after are in an "or" relationship. "At least one (one) of the following" or a similar expression means any combination of these items, including any combination of single items (ones) or plural items (ones). For example, at least one (one) of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0106] The preferred embodiments of the embodiments of this application have been described above with reference to the accompanying drawings, but this does not limit the scope of the rights of the embodiments of this application. Any modifications, equivalent replacements, and improvements made by those skilled in the art without departing from the scope and essence of the embodiments of this application shall be within the scope of the rights of the embodiments of this application.

Claims

1. A lawn mowing robot, characterized in that, The mowing robot includes: a driving module, an operation module, a navigation module, an energy module, and a vehicle frame; Among them, the driving module, the operation module, the navigation module, and the energy module are all installed on the vehicle frame; The navigation module includes two RTK antennas, a lidar, and a calculation module; The calculation module includes: A first positioning unit, configured to determine the position and orientation of the mowing robot as first positioning information by using satellite signals received by the two RTK antennas and correction data sent by a reference station; A second positioning unit, configured to, when the error of the first positioning information is greater than a first set value, construct a voxel map through the point cloud obtained by scanning with the lidar, and then determine the position and orientation of the mowing robot as second positioning information according to the voxel map; A path planning unit, configured to plan a global path according to the first positioning information or the second positioning information; A driving control unit, configured to control the driving module to drive the mowing robot to move along the global path; An operation control unit, configured to control the operation module to work for mowing.

2. The lawn mowing robot according to claim 1, characterized in that, The second positioning unit includes: A preprocessing unit, configured to determine a de-distorted point cloud by using IMU data measured by an IMU unit in the navigation module; A state estimation unit, configured to optimize by using a factor graph and register the de-distorted point cloud as a prior to perform point cloud registration; and then dynamically register the scanned point cloud to the voxel map; A point cloud matching and optimization unit, configured to update the matching between the point cloud and the voxel map by using an extended Kalman filter method to optimize the pose of the mowing robot; calculate the residual error from the point cloud to the voxel map, and minimize the error from the point cloud to the voxel map under the constraint of the IMU unit; A local map updating unit, configured to store the voxel map that is closest to the current moment and within a set time period in a sliding window manner; A loop update unit, configured to identify whether the mowing robot has ever reached the same location to eliminate cumulative errors, optimize pose estimation, and perform repositioning.

3. The lawn mowing robot according to claim 2, wherein, The loop update unit includes: A key frame management unit, configured to store the point cloud data and pose at the corresponding moment as key frames after an interval of a set period or when the pose change of the mowing robot exceeds a preset threshold; A frame search unit, configured to use a KD-Tree to search for candidate key frames similar to the current key frame among the stored key frames; A point cloud matching unit, configured to calculate the ICP matching between the candidate key frame and the current key frame to obtain a relative transformation matrix and an ICP error; A factor graph optimization unit, configured to, if the ICP error is less than a second set value, add a loop constraint to the factor graph; perform global optimization by using iSAM2 incremental optimization to correct the drift error of the voxel map.

4. A lawn mowing robot according to claim 1, characterized in that, The path planning unit includes: A boundary generation unit, configured to generate a working boundary according to the first positioning information or the second positioning information; A path generation unit for generating parallel lines at intervals of the radius of the cutter head in the working module within the working boundary, and then connecting the heads and tails of the parallel lines to form a continuous global path to completely cover the working boundary.

5. A lawn mowing robot according to claim 1, wherein, The drive control unit includes: A look-ahead point selection unit for dynamically selecting look-ahead points on the global path such that the look-ahead points maintain a set distance from the current position of the mowing robot until the look-ahead points coincide with the end point; A direction adjustment unit for dynamically adjusting the direction of the mowing robot and controlling the drive module to drive the mowing robot towards the look-ahead point.

6. The lawn mowing robot according to claim 1, characterized in that, The drive control unit further includes an obstacle avoidance unit, and the obstacle avoidance unit includes: An obstacle recognition unit for determining the object surface based on the return time of the lidar beam or the angle between the incident beam and the return beam, projecting the object surface onto the global map to identify whether the object is an obstacle; or, using the camera module in the navigation module to obtain an image, segmenting the image through an AI model, projecting the segmented object onto the global map, and then determining whether the object is an obstacle; A temporary path generation unit for generating a temporary path when an obstacle is recognized; A detour unit for, when an obstacle is recognized, expanding the boundary of the obstacle by a set range, and then controlling the mowing robot to move according to the temporary path to bypass the obstacle after the boundary expansion, and then returning to the global path.

7. A lawn mowing robot according to claim 6, characterized in that, The temporary path generation unit includes: A feasible speed space generation unit for calculating and generating the feasible speed space of the mowing robot within a dynamic time window according to the kinematic constraints of the mowing robot; A trajectory prediction unit for predicting the predicted trajectories corresponding to each speed combination within a preset time period in the feasible speed space; wherein, the speed combination includes a linear speed and an angular speed; A trajectory scoring unit for scoring the corresponding predicted trajectory according to the distance between the end point of the predicted trajectory and the target point, the forward speed of the predicted trajectory, the length of the predicted trajectory, and the distance between the predicted trajectory and the obstacle; A trajectory selection unit for determining the predicted trajectory with the highest score and then generating the temporary path from the corresponding speed combination.

8. A lawn mowing robot according to claim 6, characterized in that, The obstacle avoidance unit further includes: An obstacle boundary judgment unit for judging whether the space between the obstacle and the working boundary in the global map is greater than the width of the mowing robot to judge whether detouring the obstacle will exceed the working boundary; if detouring the obstacle will exceed the working boundary, interrupt the current path, return to the starting point of the current path, and then replan the navigation path of the remaining unworked area after subtracting the worked area.

9. A lawn mowing robot according to any one of claims 1 to 8, characterized in that, The energy module includes multiple batteries, and the battery compartments of each battery adopt a quick-release structure; The computing module further includes: An energy management unit for, if the current supplied battery has a current lower than a preset power threshold, switching the currently supplied battery to the remaining batteries and sending a message reminding to replace the battery to the management terminal of the mowing robot.

10. A control method, characterized in that, The control method includes the following steps: Determine the position and orientation of the mowing robot as the first positioning information by using the satellite signals received by two RTK antennas and the correction data sent by the reference station; When the error of the first positioning information is greater than the first set value, construct a voxel map from the point cloud obtained by lidar scanning, and then determine the position and orientation of the mowing robot as the second positioning information according to the voxel map; Plan a global path according to the first positioning information or the second positioning information; Control the driving module to drive the mowing robot to move along the global path; Control the operation module to work for mowing.

Images