Motion model establishing method and motion track planning method of wheeled walking robot

Abstract

The invention discloses a motion model building method and a motion track planning method of a wheeled walking robot, and relates to the technical field of robot navigating.The method comprises the steps that firstly, an initial motion model of the wheeled robot is built according to the obtained current walking speed of the wheeled robot in the walking state, and then the initial motion model of the wheeled robot is built according to the initial motion model; the attitude variation of the wheeled robot is obtained, the attitude variation is converted into a global coordinate system, and the motion model of the wheeled robot is constructed, so that the kinematic model of the wheeled robot can be established and obtained according to the current walking speed of the wheeled robot; therefore, the wheeled robot has the functions of intelligent walking path planning and intelligent autonomous navigation, and the wheeled robot can plan a motion model and a motion track through the method disclosed by the invention, so that the method can adapt to an area lacking an electronic map; and navigation can be carried out in an area without an electronic map.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of robot navigation technology, in particular to a wheel walking robot motion model establishment method and motion trajectory planning method. BACKGROUND

[0002] With the rapid development of intelligent navigation technology, intelligent robots are widely used in take-out, logistics distribution and industrial daily inspection processes.

[0003] In the prior art, to realize the automatic navigation of the robot, the intelligent planning of the driving trajectory of the robot is performed by importing an electronic map, and the acquisition of the electronic map is usually drawn by a staff or modeled and generated by image acquisition of the to-be-driven area through another image sensor. However, this navigation method is usually suitable for the area where the electronic map is established in advance, and for the area where the electronic map is not established or obtained, the robot lacks the ability to model the map autonomously when moving, which makes the intelligent trajectory planning and intelligent navigation autonomy of the robot poor, completely depends on the acquisition of the pre-electronic map, and cannot realize the autonomous navigation in the area lacking the electronic map. SUMMARY

[0004] The main purpose of the present application is to provide a wheel walking robot motion model establishment method and motion trajectory planning method, which aims to solve the technical problems that the intelligent trajectory planning and intelligent navigation autonomy of the robot in the prior art are poor, completely depend on the acquisition of the pre-electronic map, and the robot lacks the ability to model the map autonomously, so that it is difficult to adapt to the navigation scene lacking the electronic map.

[0005] To achieve the above purpose, in a first aspect, the present application provides a wheel walking robot motion model establishment method, comprising the following steps:

[0006] According to the obtained current walking speed of the wheel robot in the walking state, an initial motion model of the wheel walking robot is established; wherein the initial motion model is represented by formula one, and the formula one is:

[0007]

[0008] v is the current walking speed of the wheel walking robot, v x is the first speed of the wheel walking robot in the X-axis direction, v ya second partial speed of the wheeled walking robot in a Y direction, ω is an angular speed of the wheeled walking robot, θ is a yaw angle of the wheeled walking robot based on a global coordinate plane; R is a turning radius of the wheeled robot, L is a wheel spacing between any two adjacent walking wheels on the same side of the wheeled robot, and α is a steering angle of the wheeled walking robot;

[0009] According to the initial motion model, an attitude change amount of the wheeled robot is obtained; wherein the attitude change amount is expressed by Formula Two, and the Formula Two is:

[0010]

[0011] Δx is an X-axis offset amount of the wheeled robot obtained at t+1 time with a self coordinate system as a reference, Δy is a Y-axis offset amount of the wheeled robot obtained at t+1 time with the self coordinate system as the reference, and Δθ is a yaw angle change amount of the wheeled robot obtained at t+1 time with the self coordinate system as the reference;

[0012] The attitude change amount is converted to a global coordinate plane, and a motion model of the wheeled robot is constructed; wherein the motion model is expressed by Formula Three, and the Formula Three is:

[0013]

[0014] is the pose information of the wheeled robot at t+1 time, is the pose information of the wheeled robot at t time, and ∈0 is noise interference.

[0015] In an embodiment, the wheeled walking robot comprises a plurality of walking wheels which are distributed in pairs and at intervals.

[0016] Before the step of establishing the initial motion model of the wheeled walking robot according to the obtained current walking speed of the wheeled robot in the walking state, the method further comprises:

[0017] When the robot is in the walking state, for each walking wheel of the wheeled walking robot, a current motion speed of the corresponding walking wheel is collected by using an encoder; wherein the current motion speed is expressed by Formula Four, and the Formula Four is:

[0018]

[0019] v i is the current motion speed of the i-th walking wheel, N i is the number of pulse signals detected by the encoder on the i-th walking wheel within a unit time length dt, and PPRi the number of pulse signals output by the encoder on the i-th walking wheel when the walking wheel rotates one revolution, r i the radius of the i-th walking wheel;

[0020] obtaining a current walking speed of the wheeled walking robot according to the current motion speeds of all the walking wheels; wherein the current walking speed is represented by Formula Five, and the Formula Five is:

[0021]

[0022] v is the current walking speed of the wheeled walking robot, and n is the number of walking wheels of the wheeled walking robot.

[0023] In an embodiment, after the step of obtaining the current walking speed of the wheeled walking robot according to the current motion speeds of all the walking wheels, the method further comprises:

[0024] obtaining a current steering angle of the wheeled robot; wherein the current steering angle is represented by Formula Six, and the Formula Six is:

[0025]

[0026] R is the turning radius of the wheeled robot, L is the wheel spacing between any two adjacent walking wheels on the same side of the wheeled robot, and a is the steering angle of the wheeled walking robot.

[0027] Based on the same technical concept, in a second aspect, the present application further provides a motion trajectory planning method for a wheeled walking robot, the motion trajectory planning method comprising the following steps:

[0028] establishing a motion model of the wheeled walking robot by using the motion model establishment method of the first aspect;

[0029] collecting spatial data of a to-be-moved region of the robot, and establishing a spatial motion map of the to-be-moved region according to the spatial data;

[0030] planning a motion trajectory of the robot from a current position to an end position in the spatial motion map by using the motion model.

[0031] In an embodiment, the step of collecting spatial data of a to-be-moved region of the robot, and establishing a spatial motion map of the to-be-moved region according to the spatial data, comprises:

[0032] collecting a target data set of the to-be-moved region of the robot; wherein the target data set comprises point cloud data, image data of the to-be-moved region, and inertial data of the robot when collecting information in the to-be-moved region;

[0033] establishing a visual global map and a point cloud global map of the to-be-moved region respectively according to the target data set;

[0034] fusing the visual global map and the point cloud global map to obtain the spatial motion map.

[0035] In an embodiment, the step of establishing the visual global map and the point cloud global map of the to-be-moved region respectively according to the target data set comprises:

[0036] establishing a plurality of visual local maps and a plurality of point cloud local maps of the to-be-moved region according to the target data set;

[0037] performing global consistency processing on all the visual local maps according to the image processing frame data collected to establish the visual global map of the to-be-moved region;

[0038] performing global consistency processing on all the point cloud local maps according to the point cloud data collected to establish the point cloud global map of the to-be-moved region.

[0039] In an embodiment, the step of establishing the visual global map and the point cloud global map of the to-be-moved region respectively according to the target data set comprises:

[0040] analyzing the motion state of a collection device collecting the target data set according to the target data set, and establishing a plurality of visual local maps and a plurality of point cloud local maps of the to-be-moved region respectively.

[0041] In an embodiment, the step of collecting the target data set of the to-be-moved region of the robot comprises:

[0042] controlling the robot provided with an image collection device and a visual collection device to collect the target data set of the to-be-moved region.

[0043] In an embodiment, before the step of planning a motion trajectory of the robot from a current position to an end position in the spatial motion map by using the motion model, the method further comprises:

[0044] calculating a straight line motion distance of the robot from the current position to the end position in the spatial motion map;

[0045] determine a size relationship between the straight line motion distance and a preset standard distance range;

[0046] The step of planning the motion trajectory of the robot from the current position to the end position in the space motion map by using the motion model comprises:

[0047] When the straight line motion distance is within the preset standard distance range, the motion trajectory of the robot from the current position to the end position in the space motion map is planned by using the motion model, and a short distance driving trajectory of the robot is obtained; wherein the motion trajectory comprises a first motion trajectory and a second motion trajectory connected in sequence, the first motion trajectory is a Riss-Schramm curve, and the second motion trajectory is a Bezier curve.

[0048] In an embodiment, after the step of determining the size relationship between the straight line motion distance and the preset standard distance range, the step of planning the motion trajectory of the robot from the current position to the end position in the space motion map by using the motion model further comprises:

[0049] When the straight line motion distance is not within the preset standard distance range, the motion trajectory of the robot moving to the obtained stop position in the updated space motion map is planned by using the motion model, and a long distance driving trajectory of the robot is obtained; wherein the stop position is the stop position when the robot moves to the end position, and the stop position is an unknown position.

[0050] The motion model establishment method and the motion trajectory planning method of the wheeled walking robot disclosed by the application, in use, first establish an initial motion model of the wheeled robot according to the obtained current walking speed of the wheeled robot in the walking state, then obtain the attitude change amount of the wheeled robot according to the initial motion model, convert the attitude change amount to the global coordinate system, and construct the motion model of the wheeled robot, so that the application can establish and obtain the kinematic model of the wheeled robot according to the current walking speed of the wheeled robot, and the wheeled robot is also provided with the functions of intelligent planning of walking path and intelligent autonomous navigation. Since the motion model and the motion trajectory can be planned by the method disclosed by the application, the application can be adapted to the area lacking electronic maps, and can navigate in the area lacking electronic maps. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only show some of the embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from the structures shown in the drawings without creative labor.

[0052] Figure 1 The flowchart of the motion model establishment method of the wheeled walking robot provided by the present application is shown in the figure.

[0053] Figure 2 The flowchart of some specific embodiments of the motion trajectory planning method of the wheeled walking robot provided by the present application is shown in the figure. Figure 1

[0054] The flowchart of the motion trajectory planning method of the wheeled walking robot provided by the present application is shown in the figure. Figure 3

[0055] The flowchart of step A200 of the example in the present application is shown in the figure. Figure 4 Figure 3 The flowchart of step A220 of the example in the present application is shown in the figure.

[0056] Figure 5 Figure 4 The flowchart of some specific embodiments of the motion trajectory planning method of the wheeled walking robot provided by the present application is shown in the figure.

[0057] Figure 6 The flowchart of some specific embodiments of the motion trajectory planning method of the wheeled walking robot provided by the present application is shown in the figure.

[0058] The implementation, functional features and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0059] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the present application.

[0060] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back, etc.), the directional indications are only used to explain the relative position relationship, movement condition, etc. between the components in a certain posture, and if the certain posture changes, the directional indications also change accordingly.

[0061] ​​In addition, if the description of "first", "second" and the like is involved in the embodiments of the present application, the description of "first", "second" and the like is only for the purpose of description, and cannot be understood as indicating or implying the relative importance of the indicated technical features or implicitly indicating the number of the indicated technical features. Therefore, the features limited by "first", "second" can be explicitly or implicitly included at least one of the features. In addition, if "and / or" or "and / or" appears throughout the text, it means that the three parallel schemes are included, for example, "A and / or B" includes A scheme, or B scheme, or A and B scheme. In addition, the technical solutions of each embodiment can be combined with each other, but it must be based on the realization of the ordinary skilled in the art, when the combination of technical solutions appears contradictory or unachievable, it should be considered that the combination of technical solutions does not exist, nor in the protection scope required by the present application.

[0062] With the rapid development of intelligent navigation technology, intelligent robots are widely used in take-out, logistics distribution and industrial daily inspection process.

[0063] The applicant found that in the prior art, in order to realize the automatic navigation of the robot, the intelligent planning of the driving track is carried out by importing the electronic map into the robot, and the acquisition of the electronic map is usually drawn by the staff or modeled and generated by image acquisition of the to-be-driven area through another image sensor. However, this navigation method is usually suitable for the area where the electronic map is established in advance, and for the area where the electronic map is not established or obtained, the robot lacks the ability to model the map autonomously when moving, which makes the intelligent trajectory planning and intelligent navigation autonomy of the robot poor, completely depends on the acquisition of the pre-electronic map, and cannot realize the autonomous navigation in the area lacking the electronic map.

[0064] The present application provides a wheeled walking robot motion model establishment method and motion trajectory planning method.

[0065] Please refer to Figures 1 to 6 , in order to facilitate understanding, the wheeled walking robot motion model establishment method comprises the following steps:

[0066] S100, according to the current walking speed of the wheeled robot in the walking state, the initial motion model of the wheeled walking robot is established; wherein the initial motion model is represented by formula one, and the formula one is:

[0067]

[0068] v is the current walking speed of the wheeled walking robot, v x is the first speed of the wheeled walking robot in the X axis direction, v ywherein, v is a first speed component of the wheeled walking robot in a Y direction, ω is an angular velocity of the wheeled walking robot, θ is a yaw angle of the wheeled walking robot based on a global coordinate plane, R is a turning radius of the wheeled robot, L is a wheel spacing between any two adjacent walking wheels on the same side of the wheeled robot, and α is a steering angle of the wheeled walking robot.

[0069] S200, obtaining a pose change amount of the wheeled robot according to the initial motion model; wherein the pose change amount is expressed by Formula Two, and the Formula Two is:

[0070]

[0071] Δx is an X-axis offset amount of the wheeled robot at t+1 time obtained with reference to a self coordinate system, Δy is a Y-axis offset amount of the wheeled robot at t+1 time obtained with reference to the self coordinate system, and Δθ is a yaw angle change amount of the wheeled robot at t+1 time obtained with reference to the self coordinate system.

[0072] S300, converting the pose change amount to a global coordinate plane to construct a motion model of the wheeled robot; wherein the motion model is expressed by Formula Three, and the Formula Three is:

[0073]

[0074] is the pose information of the wheeled robot at t+1 time, is the pose information of the wheeled robot at t time, and ∈0 is a noise interference.

[0075] Specifically, the applicant finds that the traditional method does not fully consider the correlation between the steering geometric relationship and the global coordinate mapping by analyzing the kinematic characteristics of the wheeled robot. Based on this, the robot motion is decomposed into the cooperative calculation of the speed component and the angular velocity, the initial model containing the steering parameters is established, and the coordinate conversion from the local motion to the global space is finally realized by combining the time discretization processing. Further, the application breaks through the limitation of simply relying on the external map, and enables the robot to have the ability to solve the real-time self pose.

[0076] Therefore, the application proposes the following steps: establishing an initial motion model according to a current walking speed of a wheeled robot in a walking state, the model describing a motion state through a speed component equation and a turning geometry equation; calculating a robot pose change amount based on a time increment; and mapping local displacement to a global coordinate system through a coordinate conversion matrix to construct a motion model containing noise compensation. The initial motion model contains an X / Y axis speed component equation and an angular velocity equation, the pose change amount is obtained through the product of the speed component and the time increment, and the global coordinate conversion is realized through a rotation matrix.

[0077] More specifically, by acquiring the linear speed of the robot in real time, an initial model containing speed components and turning parameters is first established. After the linear speed at the current time is obtained, the X / Y axis components are decomposed according to the trigonometric function relationship, and the angular velocity is calculated through the geometric relationship between the wheel spacing and the turning angle. After determining the unit time increment, the speed components are converted into displacement increments, and the displacement in the local coordinate system is converted into the global coordinate system through a rotation matrix. In this process, the introduction of the noise term effectively compensates for the influence of factors such as mechanical transmission error and ground friction, making the established model closer to the actual motion state. The final motion model can continuously predict the pose change of the robot in the global coordinate system through iterative calculation.

[0078] Compared with the prior art, the traditional method relies on a pre-generated static map for trajectory planning and cannot adapt to unknown environments or dynamic obstacle scenarios. The present scheme establishes a self-contained kinematic model, so that the robot can solve the pose only by relying on its own motion parameters without relying on external environmental information. In the prior art, coordinate conversion is mostly based on simplified assumptions, ignoring the dynamic relationship between the turning parameter and the motion state, resulting in insufficient model accuracy. The present scheme accurately correlates the physical properties of the turning angle and the turning radius by introducing the turning geometry equation, significantly improving the model accuracy under complex turning conditions.

[0079] In the embodiment, first, an initial motion model of the wheeled robot is established according to the current walking speed of the wheeled robot in the walking state, and then a pose change amount of the wheeled robot is obtained according to the initial motion model, the pose change amount is converted to a global coordinate system, and a motion model of the wheeled robot is constructed, so that the application can establish and obtain the kinematic model of the wheeled robot according to the current walking speed of the wheeled robot, and the wheeled robot also has the functions of intelligent planning of a walking path and intelligent autonomous navigation. Since the wheeled robot can plan the motion model and the motion trajectory through the method disclosed in the application, the application can be adapted to areas lacking electronic maps and can navigate in areas without electronic maps.

[0080] Further, the dependence of the wheeled robot on the preset electronic map can be reduced, and the autonomous navigation function of the wheeled robot is improved.

[0081] In an embodiment, the wheeled walking robot comprises a plurality of walking wheels arranged in pairs opposite to each other and spaced apart;

[0082] Before step S100, the method further comprises:

[0083] S400, when the robot is in a walking state, for each walking wheel of the wheeled walking robot, a current motion speed of the corresponding walking wheel is collected by using an encoder; wherein the current motion speed is expressed by formula four, and the formula four is:

[0084]

[0085] v i is the current motion speed of the i th walking wheel, N i is the number of pulse signals detected by the encoder on the i th walking wheel within a unit time length dt, PPR i is the number of pulse signals output by the encoder on the i th walking wheel when the walking wheel rotates one turn, r i is the radius of the i th walking wheel.

[0086] Specifically, the encoder refers to a sensor device for measuring the rotation speed of the walking wheel, which can be implemented by using an optical or magneto-electric encoder. The wheel speed is calculated by detecting the number of pulse signals generated when the walking wheel rotates. The number of pulse signals refers to the number of electrical pulses output by the encoder within a unit time, which can be collected and processed by a counter module to reflect the number of rotation turns of the walking wheel. The physical quantity relationship in formula four converts the rotational motion into linear speed by combining the pulse signals with the wheel circumference, ensuring the accurate calculation of the speed of a single walking wheel. The mean value calculation strategy in formula five processes the speed data of all walking wheels by arithmetic mean, eliminates the measurement errors caused by the differences in ground friction or the slipping of individual wheels, and improves the reliability of the overall speed.

[0087] S500, according to the current motion speeds of all the walking wheels collected, the current walking speed of the wheeled walking robot is obtained; wherein the current walking speed is expressed by formula five, and the formula five is:

[0088]

[0089] v is the current walking speed of the wheeled walking robot, and n is the number of walking wheels of the wheeled walking robot.

[0090] Specifically, during the robot walking process, the encoder independently installed on each walking wheel detects the pulse signal generated by its rotation in real time. By recording the number of pulses in a unit of time, combining the fixed number of pulses output by the encoder per revolution and the radius of the walking wheel, the pulse signal is converted into the linear speed of each wheel by using Formula Four. Subsequently, the linear speed of all walking wheels is substituted into Formula Five for arithmetic averaging to obtain the current walking speed reflecting the overall motion state of the robot.

[0091] In the embodiment, through the dual mechanisms of multi-sensor data fusion and mean calculation, the accuracy of the speed parameter is ensured, a reliable input is provided for the subsequent motion model establishment, the dependence on the pre-electronic map is reduced, and the precision of the autonomous navigation result of the wheeled robot is improved.

[0092] In an embodiment, after step S500, the method further comprises:

[0093] S600, obtaining a current steering angle of the wheeled robot; wherein the current steering angle is represented by Formula Six, and the Formula Six is:

[0094]

[0095] R is the turning radius of the wheeled robot, L is the wheel spacing between any two adjacent walking wheels on the same side of the wheeled robot, and a is the steering angle of the wheeled robot.

[0096] Specifically, the turning radius refers to the radius of the trajectory circle when the robot turns, which can be derived through the relationship between the speed and the angular velocity in the kinematic model, and is used to represent the trajectory curvature when the robot turns. The wheel spacing refers to the straight-line distance between the center points of the adjacent walking wheels on the same side, which is obtained by pre-measuring as a fixed mechanical structure parameter, and is used to establish the geometric constraint relationship between the steering angle and the turning radius. The steering angle refers to the deflection angle of the driving wheel when the robot turns, which is related to the wheel spacing and the turning radius through the tangent function, and does not need to be directly installed with an angle sensor for indirect calculation.

[0097] On the basis of obtaining the walking speed of the robot, the fixed parameter characteristic of the wheel spacing between the adjacent wheels on the same side is utilized, the dynamic change value of the turning radius is combined, and the steering angle is converted into a calculable algebraic parameter through the geometric relationship tan a = L / R. For example, when the robot performs a turning action, the turning radius R changes in real time with the motion state, at this time, based on the known wheel spacing L, the value of the steering angle a can be directly calculated through the tangent function relationship. The calculation of the steering control parameter is embedded in the kinematic model in the manner of step S600, avoiding the hardware complexity and error accumulation problem introduced by additional sensors, while ensuring that the steering angle parameter is strictly matched with the mechanical structure characteristics, thereby improving the accuracy of the turning trajectory control.

[0098] Compared with the prior art, the traditional method usually relies on a steering angle sensor to directly measure the physical deflection of a mechanical component, but the sensor installation accuracy is easily affected by mechanical vibration or environmental interference, resulting in measurement error. However, the present scheme converts the steering angle into a derived quantity based on structural parameters and motion parameters through the geometric constraint relationship inherent in the kinematic model, without relying on external sensors, thereby reducing hardware costs and improving the stability and reliability of parameter calculation.

[0099] Based on the same technical concept, in a second aspect, please continue to refer to Figures 3 to 6 The present application also provides a motion trajectory planning method for a wheeled walking robot, which comprises the following steps:

[0100] A100, using the motion model establishment method of the first aspect to establish the motion model of the wheeled walking robot;

[0101] A200, collecting spatial data of a region to be moved of the robot, and establishing a spatial motion map of the region to be moved according to the spatial data;

[0102] A300, planning a motion trajectory of the robot from a current position to an end position in the spatial motion map by using the motion model.

[0103] Specifically, when the robot enters an unknown region, first, the current walking speed and steering angle are calculated in real time based on the wheel speed data collected by the encoder, and a kinematic model containing the relationship between the speed component and the angular velocity is established. Then, the environment point cloud data, image data and inertial measurement data are synchronously collected by the visual sensor and the laser radar, and a visual local map and a point cloud local map are respectively constructed, and a globally consistent spatial motion map is generated through coordinate alignment and data fusion. Finally, the pose change equation in the motion model and the obstacle distribution information in the spatial map are input into the trajectory planning algorithm, and the optimal path that satisfies the kinematic constraint and avoids obstacles is generated through iterative optimization.

[0104] In this embodiment, the environment map is autonomously constructed by real-time collection of multi-modal perception data, and dynamic trajectory planning is realized in combination with the kinematic model, thereby solving the navigation problem in the case of missing pre-set map. The map constructed by a single sensor in the prior art is easily disturbed by noise, resulting in failure of trajectory planning. The present scheme effectively improves the robustness of environmental representation through fusion processing of visual and point cloud data. At the same time, a high-precision environment map is constructed through multi-source data fusion, ensuring accurate perception of obstacle distribution during the trajectory planning process. The path generated in combination with the kinematic model not only conforms to the steering characteristics of the robot, but also can automatically select a short-distance or long-distance optimization strategy according to the complexity of the environment, thereby improving the navigation efficiency and safety.

[0105] In an embodiment, step A200 comprises:

[0106] A210, collecting a target data set of the to-be-moved region of the robot; wherein the target data set comprises point cloud data, image data of the to-be-moved region, and inertial data of the robot when collecting information in the to-be-moved region;

[0107] A220, respectively establishing a visual global map and a point cloud global map of the to-be-moved region according to the target data set;

[0108] A230, fusing the visual global map and the point cloud global map to obtain the spatial motion map.

[0109] Specifically, by collecting a target data set containing point cloud, image and inertial data, the point cloud data provides three-dimensional spatial structure information, the image data supplements texture details, and the inertial data corrects the pose error in the movement process. Then, a visual global map and a point cloud global map are respectively constructed. The visual map establishes an appearance model of the scene through visual feature matching, and the point cloud map establishes a geometric model of the scene through point cloud registration. Finally, through fusion processing, the two types of maps are aligned and redundant information is eliminated, combining the scene recognition ability of visual information and the spatial precision of point cloud data to form a unified spatial motion map. In this embodiment, the inertial data is used to compensate for the pose deviation caused by bumps or sliding during robot movement, ensuring the continuity of map construction.

[0110] In this embodiment, through real-time collection and fusion of multi-source data, a high-precision map can be constructed autonomously in an unknown environment without relying on a pre-stored map. In the prior art, the visual map is easily disturbed by changes in light, and the point cloud map lacks semantic information. The present scheme complements the geometric precision and scene recognition ability by fusing the two types of maps.

[0111] In an embodiment, step A220 comprises:

[0112] A221, establishing a plurality of visual local maps and a plurality of point cloud local maps of the to-be-moved region according to the target data set;

[0113] A222, performing global consistency processing on all the visual local maps according to the image processing frame data obtained, to establish the visual global map of the to-be-moved region;

[0114] A223, performing global consistency processing on all the point cloud local maps according to the point cloud data obtained, to establish the point cloud global map of the to-be-moved region.

[0115] Specifically, in the establishment of the visual local map, by analyzing the image data in the target data set, the feature points in each frame of image are extracted and their spatial poses are calculated, thereby generating a plurality of local maps covering different areas. Subsequently, the time sequence information in the image processing frame data is utilized to detect the overlapping feature points between adjacent visual local maps, and the plurality of local maps are spliced into a visual global map through optimization of the pose transformation matrix. For the point cloud local map, by matching the point cloud features in different local maps, such as planar or edge structures, the rotation and translation parameters are calculated to realize point cloud registration, and finally fused into a point cloud global map.

[0116] In the present embodiment, a hierarchical processing mechanism is adopted, the visual and point cloud local maps are first independently constructed to preserve the sensor characteristics, and then the local errors are eliminated through multi-source data fusion, such as scale drift of the visual map and noise interference of the point cloud map. By separating the visual and point cloud data processing procedures, the time sequence correlation of the image frames and the spatial structure characteristics of the point cloud are utilized for optimization, which not only avoids the errors introduced by the coupling of the sensor data, but also improves the global consistency through the complementarity of the multi-modal data.

[0117] In an embodiment, step A221 comprises:

[0118] According to the target data set, the motion state of the acquisition device for acquiring the target data set is analyzed, and a plurality of visual local maps and a plurality of point cloud local maps of the target motion region are established respectively.

[0119] Specifically, the establishment of the visual local map refers to extracting environmental feature points based on image data, generating two-dimensional or three-dimensional visual maps through feature matching algorithm, which can be realized by ORB feature detection and SLAM technology. This step preserves the environmental texture information and provides high-resolution visual features for the subsequent global map. The establishment of the point cloud local map refers to acquiring three-dimensional space point cloud data through laser radar or depth camera, and performing point cloud registration through ICP algorithm, which can be realized by voxel filtering denoising processing. This step provides accurate spatial geometric structure and makes up for the deficiency of visual data in depth information.

[0120] When acquiring the target data set, the motion state parameters of the device are recorded synchronously, including linear velocity, angular velocity and attitude angle. For visual data, the exposure time offset during image acquisition is compensated according to the motion state of the device, and the motion blur is eliminated; for point cloud data, the point cloud is coordinate-transformed according to the device pose change, and the point cloud stretching distortion caused by device movement is eliminated. Through the time stamp alignment mechanism, the visual frame data and the point cloud data are bound to the same motion state node, and the local maps are constructed in their respective coordinate systems. The visual local map generates a topological structure through feature point clustering, and the point cloud local map extracts geometric boundaries through plane segmentation. Both of them are independently stored in the spatial coordinate system but maintain a time synchronization relationship.

[0121] In the present embodiment, dynamic compensation of the data acquisition process is achieved through motion state perception, and visual and point cloud data are corrected for motion errors during the construction of local maps, avoiding error accumulation and amplification in subsequent fusion stages. In the prior art, when a single visual or point cloud data is used for mapping, the visual map is easily disturbed by changes in light, and the point cloud map lacks semantic information. The present scheme separates the construction of two local maps, retains the advantages of each data, and provides complementary data basis for global map fusion.

[0122] In an embodiment, step A210 comprises:

[0123] Controlling the robot equipped with the image acquisition device and the visual acquisition device to acquire the target data set of the region to be moved.

[0124] Specifically, the image acquisition device refers to a device for obtaining two-dimensional image information of the environment, which can be implemented by a camera or an optical sensor. Its function is to capture the texture, color and planar features of the environment. The visual acquisition device refers to a device for obtaining three-dimensional spatial information, which can be implemented by a laser radar or a depth camera. Its function is to provide point cloud data and depth information of the environment. The target data set refers to a set containing multi-modal perception data of the region to be moved, which can include point cloud data, image data and inertial measurement data. Its function is to provide original input for map construction. The robot equipped with the image acquisition device and the visual acquisition device refers to an integrated system formed by physically connecting the sensor and the robot body. It can be implemented by rigid fixation or calibration. Its function is to ensure that the sensor data and the robot motion state are synchronized in space and time.

[0125] During the movement of the robot, the image acquisition device captures environmental images at a preset frequency, such as 30 frames of RGB images per second, while the visual acquisition device synchronously acquires three-dimensional point cloud data. The data of the image acquisition device and the visual acquisition device are aligned by timestamp to form the target data set, in which the image data provides planar feature information and the point cloud data supplements spatial geometric structure. Since the sensor and the robot body are directly integrated, no external equipment or manual intervention is required during the acquisition process, avoiding data delay or calibration errors caused by separation of devices. For example, when the robot moves to an unknown area, the image acquisition device acquires wall texture information in real time, and the visual acquisition device records obstacle distance data synchronously. The combination of the two can be directly used for subsequent map construction.

[0126] In this embodiment, the robot autonomously collects multi-source data, dynamically updates environmental information during movement, does not need to obtain a map in advance, and thus can improve the autonomous navigation ability of the robot in an unknown environment. A dynamic map is constructed by real-time collection of multi-modal data, solving the problem of dependence on a pre-stored map in traditional methods. At the same time, the integrated design of the sensor and the robot body avoids data synchronization error, ensures the spatio-temporal consistency of environmental perception and movement state, and provides a reliable foundation for subsequent trajectory planning.

[0127] In an embodiment, before step A300, the method further comprises:

[0128] A400, calculating a straight-line movement distance of the robot from the current position to the end position in the spatial motion map;

[0129] A500, judging the size relationship between the straight-line movement distance and a preset standard distance range;

[0130] Step A300 comprises:

[0131] A310, when the straight-line movement distance is within the preset standard distance range, a motion trajectory of the robot from the current position to the end position in the spatial motion map is planned using the motion model, to obtain a short-distance driving trajectory of the robot; wherein the motion trajectory comprises a first motion trajectory and a second motion trajectory connected in sequence, the first motion trajectory is a Reuleaux-Scheuer curve, and the second motion trajectory is a Bezier curve.

[0132] Specifically, the straight-line movement distance refers to the Euclidean distance between the current position and the target position of the robot, which can be calculated using a coordinate point distance formula and is used to quantify the size of the trajectory planning task. The preset standard distance range refers to a threshold parameter for dividing short-distance and long-distance scenarios, which can be obtained by experimental testing of the critical value in a typical scenario and is used to trigger different trajectory planning strategies. The Reuleaux-Scheuer curve refers to a parameterized curve that satisfies the second-order continuous derivative characteristic, which can be constructed using a cubic polynomial function, and its curvature continuity characteristic can avoid energy loss caused by path mutation. The minimum energy loss constraint condition refers to the conversion of the power consumption model of the driving motor into a functional relationship between path curvature and speed, which can be realized by establishing a mapping relationship between energy consumption and movement parameters, and is used to optimize the economy of the path.

[0133] When the navigation system detects that the target point distance exceeds the preset threshold, the trajectory planning module starts the energy optimization mode. The robot's angular velocity and linear velocity combinations under different paths are calculated through the motion model, and a candidate path set is generated in combination with the mathematical properties of the Reeds-Shepp curve. In the path evaluation stage, the energy consumption model is used to calculate the driving system power consumption corresponding to each path, and the path that meets the kinematic constraints and has the lowest energy consumption is selected as the final solution. In this process, the continuous curvature property of the Reeds-Shepp curve ensures smooth transition of the robot's turning action, avoiding sudden turns that cause sudden changes in motor load, and the introduction of energy constraints ensures that the path selection always follows the principle of minimum power consumption.

[0134] It needs to be particularly and explicitly pointed out that in the embodiment, the example of the preset standard distance range is preferably 3m.

[0135] In an embodiment, after step A500, step A300 further comprises:

[0136] A320, when the straight line motion distance is not located in the preset standard distance range, the motion trajectory of the robot moving to the obtained stop position is planned in the updated spatial motion map using the motion model, and the long-distance driving trajectory of the robot is obtained; wherein the stop position is the stop position when the robot moves to the end position, and the stop position is an unknown position.

[0137] Specifically, in the embodiment, when the execution motion distance is not located in the preset standard distance range, the stop position (i.e. the end position) of the robot needs to be determined first. When determining the stop position, it should be judged whether the stop position is known. When the stop position is known, the Hybrid A* algorithm is directly used in combination with the dynamic window algorithm to plan the motion trajectory of the wheeled robot moving to the stop position. When the stop position is unknown, the sensor module arranged on the robot is used to collect the sensing data (the sensing data at least includes image data of the area where the robot is located) of the area where the robot is located, then the new global point cloud map is obtained based on the collected sensing data in the manner of step A200, and the frontier exploration algorithm is used to obtain the exploration target point. After determining the exploration target point, it is judged whether the stop position is known. When the target point is not the stop position, i.e. the stop position is unknown, the Hybrid A* algorithm is used in combination with the dynamic window algorithm to plan the motion trajectory of the wheeled robot moving to the exploration target point. When the stop position is determined, i.e. the stop position is known, the Hybrid A* algorithm is used in combination with the dynamic window algorithm to plan the motion trajectory of the wheeled robot moving to the stop position.

[0138] It needs to be made clear that in the present embodiment, the example adoption of the frontier exploration algorithm combined with the dynamic window algorithm plans the overall motion trajectory of the wheeled robot and the manner of obtaining the stop position in the case where the stop position is unknown are prior art, which are only applied in the present embodiment and are not improved in design, and therefore the specific process is not described in detail here.

[0139] In the above embodiment, the wheeled walking robot can plan an optimal long-distance motion trajectory in terms of energy consumption in the scenario without a pre-existing electronic map, and the mechanical loss of the steering mechanism is reduced through a smooth path, thereby prolonging the service life of the equipment. Meanwhile, the energy waste problem caused by the sudden change of path curvature in the traditional method is solved, and the autonomous navigation efficiency of the robot in an unknown environment is improved.

[0140] Of course, in other specific embodiments, when implementing the example scheme of the present application, it is necessary to first establish a motion model of the wheeled robot according to the current walking speed of the wheeled robot obtained by acquisition, and also to establish a geometric model of the wheeled robot. While establishing the kinematic model, the geological data of the area to be walked by the wheeled robot is synchronously acquired by the wheeled robot, and a corresponding global static coordinate system, an odometer coordinate system and the projection center of the wheeled robot chassis on the ground are established according to the geological data of the corresponding area, and the transformation relationship between the coordinate systems is obtained. After obtaining the global static coordinate system, the odometer coordinate system, the projection center of the wheeled robot chassis on the ground and the transformation relationship between the coordinate systems, the SLAM algorithm is used to fuse the visual data and the IMU data obtained by the inertial measurement unit to realize high-precision positioning through the mapping update mechanism. On the basis of realizing high-precision positioning, the initial position of the wheeled robot is recorded based on the txt file, and after completing the initial position recording, the YOLO algorithm is used to recognize the image data in the area where the wheeled robot is located within a preset distance range, and the navigation end point of the wheeled robot is determined according to the image data. After determining the navigation end point, it is determined whether the current distance between the end point and the wheeled robot is within the preset standard distance range. If yes, the trajectory planning is performed in a combination and transition manner of the Riss-Schramm curve, the third-order Bezier curve and the straight line trajectory, and then the motion trajectory of the wheeled robot from the current position to the end position within the preset standard distance range is obtained. When the current distance between the end point and the wheeled robot is not within the preset standard distance range, the motion trajectory of the wheeled robot from the current position to the end position is planned in a combination of global planning and local planning. Here, there are two cases. One is when the stop position is unknown, the overall motion trajectory of the wheeled robot can be planned by using the front exploration algorithm combined with the Hybrid A* algorithm when global planning is performed. After completing the overall motion trajectory planning of the wheeled robot, the local motion trajectory planning of the wheeled robot can be performed, and the dynamic window algorithm can be used when planning the local motion trajectory. The other is when the stop position is known, the overall motion trajectory of the wheeled robot can be planned by using the Hybrid A* algorithm when global planning is performed. After completing the overall motion trajectory planning of the wheeled robot, the local motion trajectory planning of the wheeled robot can be performed, and the dynamic window algorithm can be used when planning the local motion trajectory.

[0141] Of course, in the present embodiment, the use of the dynamic window algorithm to plan the motion trajectory of the wheeled robot is an example of the prior art, which will not be described here.

[0142] The above merely illustrates the embodiments of the present application, and is not intended to limit the patent scope of the present application. Any equivalent structural transformation, direct / indirect application in other related technical fields, or the like, within the technical concept of the present application, and based on the content of the present application and the accompanying drawings, are included in the patent protection scope of the present application.

Claims

1. A method for establishing a motion model of a wheeled walking robot, characterized by, Comprising the following steps: According to the obtained current walking speed of the wheeled robot in the walking state, an initial motion model of the wheeled walking robot is established; wherein the initial motion model is represented by formula one, and the formula one is: v is the current walking speed of the wheeled walking robot, v x v is the first component speed of the wheeled walking robot in the X-axis direction, v y v is the second component speed of the wheeled walking robot in the Y direction, ω is the angular speed of the wheeled walking robot, θ is the yaw angle of the wheeled walking robot based on the global coordinate plane; R is the turning radius of the wheeled robot, L is the wheel spacing between any two adjacent walking wheels on the same side of the wheeled robot, and α is the steering angle of the wheeled walking robot. According to the initial motion model, the attitude change amount of the wheeled robot is obtained; wherein the attitude change amount is represented by formula two, and the formula two is: Δx is the X-axis offset amount of the wheeled robot at t+1 time obtained with its own coordinate system as reference, Δy is the Y-axis offset amount of the wheeled robot at t+1 time obtained with its own coordinate system as reference, and Δθ is the yaw angle change amount of the wheeled robot at t+1 time obtained with its own coordinate system as reference; The attitude change amount is converted to a global coordinate plane to construct a motion model of the wheeled robot; wherein the motion model is represented by formula three, and the formula three is: the pose information of the wheeled robot at time t+1, the pose information of the wheeled robot at time t, and ∈0is a noise disturbance.

2. The method of claim 1, wherein the motion model is established based on a plurality of motion data of the wheeled mobile robot. The wheeled walking robot comprises a plurality of walking wheels which are oppositely arranged and spaced apart; Before the step of establishing the initial motion model of the wheeled walking robot according to the obtained current walking speed of the wheeled robot in the walking state, the method further comprises: When the robot is in the walking state, the current motion speed of each walking wheel of the wheeled walking robot is collected by using an encoder; wherein the current motion speed is represented by formula four, and the formula four is: v i V is the current motion speed of the i-th walking wheel, N i PPR is the number of pulse signals detected by the encoder on the i-th walking wheel in a unit time dt i r is the number of pulse signals output by the encoder on the i-th walking wheel when the walking wheel rotates one revolution i Ri is the radius of the i-th walking wheel According to the collected current motion speeds of all the walking wheels, the current walking speed of the wheeled walking robot is obtained; wherein the current walking speed is represented by formula five, and the formula five is: v is the current walking speed of the wheeled walking robot, and n is the number of walking wheels of the wheeled walking robot.

3. The method of claim 2, wherein the motion model is established based on a plurality of motion data of the wheeled mobile robot. After the step of obtaining the current walking speed of the wheeled walking robot according to the collected current motion speeds of all the walking wheels, the method further comprises: The current steering angle of the wheeled robot is obtained; wherein the current steering angle is represented by formula six, and the formula six is: R is the turning radius of the wheeled robot, L is the wheel spacing between any two adjacent walking wheels on the same side of the wheeled robot, and α is the steering angle of the wheeled walking robot.

4. A method for planning a motion trajectory of a wheeled walking robot, characterized by, The motion trajectory planning method comprises the following steps: The motion model of the wheeled walking robot is established by using the motion model establishment method according to any one of claims 1 to 3; The spatial data of the motion region of the robot is collected, and a spatial motion map of the motion region is established according to the spatial data; The motion trajectory of the robot from the current position to the end position is planned in the spatial motion map by using the motion model.

5. The method of Claim 4, wherein, The step of collecting the spatial data of the motion region of the robot and establishing a spatial motion map of the motion region according to the spatial data comprises: Collect a target data set of the to-be-moved region of the robot; wherein the target data set comprises point cloud data, image data of the to-be-moved region, and inertial data of the robot when collecting information in the to-be-moved region; According to the target data set, a visual global map and a point cloud global map of the to-be-moved region are respectively established; The visual global map and the point cloud global map are fused to obtain the spatial motion map.

6. The method of Claim 5, wherein, The step of establishing the visual global map and the point cloud global map of the to-be-moved region according to the target data set comprises: According to the target data set, a plurality of visual local maps and a plurality of point cloud local maps of the to-be-moved region are established; According to the image processing frame data collected, all the visual local maps are globally consistent, and the visual global map of the to-be-moved region is established; According to the point cloud data collected, all the point cloud local maps are globally consistent, and the point cloud global map of the to-be-moved region is established.

7. The method of Claim 6, wherein, The step of establishing the visual global map and the point cloud global map of the to-be-moved region according to the target data set comprises: According to the target data set, the motion state of the collection device collecting the target data set is analyzed, and a plurality of visual local maps and a plurality of point cloud local maps of the to-be-moved region are respectively established.

8. The method of Claim 5, wherein, The step of collecting the target data set of the to-be-moved region of the robot comprises: The robot equipped with an image collection device and a visual collection device is controlled to collect the target data set of the to-be-moved region.

9. The method of Claim 4, wherein, Before the step of planning a motion trajectory of the robot from a current position to an end position in the spatial motion map by using the motion model, the method further comprises: Calculating a straight line motion distance of the robot from the current position to the end position in the spatial motion map; Judging the size relationship between the straight line motion distance and a preset standard distance range; The step of planning a motion trajectory of the robot from a current position to an end position in the spatial motion map by using the motion model comprises: When the straight line motion distance is within the preset standard distance range, a motion trajectory of the robot from the current position to the end position in the spatial motion map is planned by using the motion model, and a short distance driving trajectory of the robot is obtained; wherein the motion trajectory comprises a first motion trajectory and a second motion trajectory connected in sequence, the first motion trajectory is a Reis-Schramm curve, and the second motion trajectory is a Bezier curve.

10. The method of Claim 9, wherein, After the step of judging the size relationship between the straight line motion distance and the preset standard distance range, the step of planning a motion trajectory of the robot from a current position to an end position in the spatial motion map by using the motion model further comprises: When the straight line motion distance is not within the preset standard distance range, a motion trajectory of the robot moving to a obtained stop position is planned in the updated spatial motion map by using the motion model, and a long distance driving trajectory of the robot is obtained; wherein the stop position is a stop position when the robot moves to an end position, and the stop position is an unknown position.

Images