An Inspection and Positioning Method and System for an Intelligent Bionic Robot at a Grain Terminal

Abstract

This application relates to the technical field of inspection and positioning, specifically a method and system for an intelligent bionic robot to inspect and position at a grain terminal. In this application, by adjusting the foot-end contact angle according to the ground characteristics and foot pressure values, the robot can be adjusted according to the real-time ground feedback, optimizing the contact angle and posture. By measuring the bottom contact pressure and slip adjustment value of the foot in real time, the robot can make dynamic adjustments according to the actual situation of each step, reducing slip caused by changes in ground friction or obstacles. By obtaining and analyzing multiple angular accelerations and linear accelerations in real time and combining with the local motion trajectory, more accurate motion state information can be provided for the robot. By obtaining three-dimensional environmental images in each local motion trajectory, a corresponding local optimized map can be formed. The robot can flexibly adjust its own motion path and foot-end contact angle according to these optimized maps, thereby optimizing the inspection path and improving the overall inspection efficiency and safety.

Description

Technical Field

[0001] The present invention relates to the technical field of inspection and positioning, and particularly relates to a method and system for inspecting and positioning an intelligent bionic robot at a grain terminal. Background Art

[0002] Bionic robots are a type of robot that mimics the movement mechanisms of organisms. By imitating the movement structures and behaviors of animals or humans in nature, their adaptability and efficiency in complex environments are improved. Bionic robots usually have flexible movement methods, such as imitating the gait of quadruped animals or the movements of human arms and legs, making them highly adaptable in complex terrains and dynamic environments. A grain terminal is a complex and challenging working environment, the ground of which is usually composed of piled grains, goods, sand, slippery cement, steel, or other industrial facilities, with uneven terrain and may be affected by different weather conditions. To ensure production efficiency and safety, grain terminals usually require regular inspections and maintenance to check the grain stacking, equipment operation, environmental sanitation, etc.

[0003] Currently, when bionic robots are in the inspection process, they usually optimize the trajectory by calculating the error between the current position and the target position. However, grain terminals usually have obstacles such as uneven ground, scattered items, and piled goods. These factors cause the ground to have undulating heights and may have uneven friction on the surface. The existing bionic robot inspection and positioning technology lacks adjustment of the foot-end contact angle, which causes the robot to be unable to adapt to the ground changes in a timely manner when facing irregular terrains, resulting in positioning errors, and further causing the bionic robot to miss some areas that need to be inspected, thus affecting the efficiency and quality of the inspection. Summary of the Invention

[0004] The main object of the present invention is to provide a method for inspecting and positioning an intelligent bionic robot at a grain terminal, aiming to solve the technical problems in the prior art.

[0005] The present invention proposes a method for inspecting and positioning an intelligent bionic robot at a grain terminal, including:

[0006] Obtain the ground features of multiple preset sections of the grain terminal and the initial foot-end joint angles of the bionic robot, and adjust each initial foot-end joint angle of the bionic robot according to each initial foot-end joint angle and each ground feature;

[0007] Obtain the position features of the bionic robot and multiple foot bottom contact pressure values after adjusting each initial foot-end joint angle, and obtain the corresponding slip adjustment values according to the position features and multiple foot bottom contact pressure values;

[0008] Adjust the foot contact points of the bionic robot according to each of the slip adjustment values, and obtain in real time the multiple angular accelerations and linear accelerations of the bionic robot within each preset section during the adjustment process;

[0009] Generate the local motion trajectories of the bionic robot within each preset section according to the multiple angular accelerations and linear accelerations, and obtain the three-dimensional environmental images of the grain terminal at multiple consecutive moments within each local motion trajectory;

[0010] Obtain the corresponding local environmental maps according to the multiple three-dimensional environmental images, and obtain the corresponding local optimized maps according to each of the local environmental maps;

[0011] Obtain the corresponding total point cloud data sets according to each of the local motion trajectories, and generate a three-dimensional positioning inspection model according to the multiple total point cloud data sets and the local optimized maps, so that the bionic robot performs inspection positioning according to the three-dimensional positioning inspection model.

[0012] Preferably, the step of adjusting each initial foot end joint angle of the bionic robot according to each of the initial foot end joint angles and each ground feature includes:

[0013] Obtain the height data of multiple adjacent points according to each of the ground features, and obtain the corresponding ground normal vector according to the multiple height data;

[0014] Obtain the corresponding ground horizontal slope and ground vertical slope according to each of the ground normal vectors, and obtain the corresponding ground slope angle according to each of the ground horizontal slope and ground vertical slope;

[0015] Obtain the corresponding foot end contact plane normal vector of the bionic robot according to each of the initial foot end joint angles, and obtain the corresponding initial foot end grounding angle according to each of the foot end contact plane normal vectors and the ground normal vectors;

[0016] Judge whether each of the initial foot end grounding angles is different from the ground slope angle;

[0017] If the initial foot end grounding angle is different from the ground slope angle, obtain the corresponding error angle according to each of the initial foot end grounding angles and the ground slope angle, and adjust the initial foot end joint angle of the bionic robot according to each of the error angles.

[0018] Preferably, the step of obtaining the slip adjustment value according to the position feature and multiple foot bottom contact pressure values includes:

[0019] Obtain the bionic robot's center of gravity position information and each foot end position information according to the position feature, and obtain the corresponding foot weight vector according to the center of gravity position information and each foot end position information;

[0020] Obtain the torque vector sum based on multiple foot touchdown pressure values and foot weight vectors, and determine whether the torque vector sum is zero;

[0021] If the torque vector sum is zero, it is determined that the bionic robot does not produce slip;

[0022] If the torque vector sum is not zero, it is determined that the bionic robot produces slip, and obtain the reference force, friction coefficient, real-time foot end contact angle, and motion characteristics of the bionic robot at this time, where the motion characteristics include real-time speed, real-time acceleration, total mass, and gravitational acceleration;

[0023] Obtain the weight of the bionic robot according to the total mass and gravitational acceleration, and obtain the vertical normal force according to the weight of the bionic robot and the real-time foot end contact angle;

[0024] Obtain the inertial force according to the total mass and real-time acceleration, and obtain the horizontal normal force according to the inertial force and the real-time foot end contact angle;

[0025] Obtain the total normal force according to the horizontal normal force and the vertical normal force, and obtain the frictional force according to the total normal force and the friction coefficient;

[0026] Obtain the slip adjustment factor according to the reference force and the frictional force, and obtain the slip adjustment value according to the slip adjustment factor, real-time speed, and real-time acceleration.

[0027] Preferably, the step of generating a local motion trajectory according to multiple angular accelerations and linear accelerations includes:

[0028] Obtain the sampling frequency and initial characteristics of the bionic robot, where the initial characteristics include initial position information, initial linear velocity, initial angular velocity, and initial angle;

[0029] Obtain the time step according to the sampling frequency, and obtain the corresponding real-time updated linear velocity according to the time step, initial linear velocity, and each linear acceleration;

[0030] Obtain the real-time updated position according to each real-time updated linear velocity, initial position, linear acceleration, and time step;

[0031] Obtain the corresponding real-time updated angular velocity according to the time step, initial angular velocity, and each angular acceleration;

[0032] Obtain the real-time updated angle according to each real-time updated angular velocity, angular acceleration, initial angle, and time step;

[0033] Taking time as the X-axis, angle as the Y-axis, and position as the Z-axis, a three-dimensional time-angle-position coordinate axis is established, and the initial position information and initial angular velocity are plotted as the starting point on the three-dimensional time-angle-position coordinate axis;

[0034] Taking the real-time updated angle and real-time updated position corresponding to each time step as connection points and plotting them on the three-dimensional time-angle-position coordinate axis;

[0035] Connecting the starting point and multiple connection points in sequence through curves to obtain a local motion trajectory.

[0036] Preferably, the step of obtaining a local environment map based on multiple three-dimensional environment images and obtaining a local optimized map based on the local environment map includes:

[0037] Dividing each three-dimensional environment image into multiple local windows and obtaining multiple pixel points according to each local window;

[0038] Obtaining the horizontal gradient and vertical gradient of each pixel point, and obtaining the sum of the squares of the horizontal gradients according to multiple horizontal gradients;

[0039] Obtaining the sum of the squares of the vertical gradients according to multiple vertical gradients, and obtaining the sum of the products of the gradients according to multiple vertical gradients and horizontal gradients;

[0040] Generating an autocorrelation matrix according to the sum of the products of the gradients, the sum of the squares of the horizontal gradients, and the sum of the squares of the vertical gradients, and obtaining the determinant value and trace value according to the autocorrelation matrix;

[0041] Obtaining a response function value according to the determinant value and trace value, and determining whether the response function value is greater than a preset threshold;

[0042] If the response function value is greater than the preset threshold, it is determined that the pixel point is a corner point;

[0043] Obtaining corresponding matching points according to multiple corner points, and constructing a local environment map according to multiple matching points by using the SLAM algorithm;

[0044] Extracting feature points corresponding to each corner point according to the local environment map, and obtaining a reprojection error according to multiple feature points and corner points;

[0045] Optimally adjusting the feature points in the local environment map according to the reprojection error to obtain a local optimized map.

[0046] Preferably, the step of obtaining a corresponding total point cloud data set according to each local motion trajectory and generating a three-dimensional positioning and inspection model according to multiple total point cloud data sets and the local optimized map includes:

[0047] Collect the point cloud data sets to be registered at multiple preset positions of the grain terminal in real time according to each of the local motion trajectories, and select the point cloud data set at the central position as the target point cloud data set;

[0048] For each point to be registered in each of the point cloud data sets to be registered, obtain the nearest neighbor matching point from the target point cloud data set using the Euclidean distance metric method;

[0049] Obtain the distance difference according to each of the nearest neighbor matching points and the points to be registered, and adjust the positions of the points to be registered in each of the point cloud data sets to be registered according to the distance difference to obtain the corresponding registered point cloud data sets;

[0050] Convert each registered point cloud data set to the global coordinate system and delete the duplicate registered points to obtain the total point cloud data set;

[0051] Use the PSR surface reconstruction algorithm to convert each total point cloud data set and the corresponding local optimization map into a local three-dimensional inspection model;

[0052] Obtain the edge coincidence points of every two adjacent local three-dimensional inspection models, and combine and splice the multiple local three-dimensional inspection models according to each of the edge coincidence points to obtain a three-dimensional positioning inspection model.

[0053] This application also provides an intelligent bionic robot inspection and positioning system in a grain terminal, including:

[0054] A first adjustment module, configured to obtain the ground features of multiple preset sections of the grain terminal and the initial foot end joint angles of the bionic robot, and adjust each of the initial foot end joint angles of the bionic robot according to each of the initial foot end joint angles and each ground feature;

[0055] A first acquisition module, configured to obtain the position features of the bionic robot and multiple foot bottom contact pressure values after adjusting each of the initial foot end joint angles, and obtain the corresponding slip adjustment values according to the position features and the multiple foot bottom contact pressure values;

[0056] A second acquisition module, configured to adjust the foot contact points of the bionic robot according to each of the slip adjustment values, and obtain multiple angular accelerations and linear accelerations of the bionic robot in each preset section in real time during the adjustment process;

[0057] A first generation module, configured to generate the local motion trajectories of the bionic robot in each preset section according to the multiple angular accelerations and linear accelerations, and obtain the three-dimensional environmental images of the grain terminal at multiple consecutive moments in each local motion trajectory;

[0058] A third acquisition module, configured to obtain a corresponding local environment map according to multiple three-dimensional environment images, and obtain a corresponding local optimized map according to each local environment map;

[0059] A second generation module, configured to obtain a corresponding total point cloud data set according to each local motion trajectory, and generate a three-dimensional positioning and inspection model according to multiple total point cloud data sets and local optimized maps, so that the bionic robot performs inspection and positioning according to the three-dimensional positioning and inspection model.

[0060] Preferably, the first adjustment module includes:

[0061] A first acquisition unit, configured to obtain height data of multiple adjacent points according to each ground feature, and obtain a corresponding ground normal vector according to the multiple height data;

[0062] A second acquisition unit, configured to obtain a corresponding ground horizontal slope and a ground vertical slope according to each ground normal vector, and obtain a corresponding ground slope angle according to each ground horizontal slope and ground vertical slope;

[0063] A third acquisition unit, configured to obtain a corresponding foot end contact plane normal vector of the bionic robot according to each initial foot end joint angle, and obtain a corresponding initial foot end grounding angle according to each foot end contact plane normal vector and the ground normal vector;

[0064] A judgment unit, configured to judge whether each initial foot end grounding angle is different from the ground slope angle;

[0065] If the initial foot end grounding angle is different from the ground slope angle, then obtain a corresponding error angle according to each initial foot end grounding angle and the ground slope angle;

[0066] An adjustment unit, configured to adjust the initial foot end joint angle of the bionic robot according to each error angle.

[0067] The present invention also provides a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the steps of the above-mentioned method for inspecting and positioning a bionic robot at a grain terminal are implemented.

[0068] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned method for inspecting and positioning a bionic robot at a grain terminal are implemented.

[0069] The beneficial effects of the present invention are as follows: By dynamically adjusting each initial foot end joint angle, the bionic robot can better adapt to these irregular terrains. By adjusting the foot end contact angle according to the ground characteristics and foot pressure values, the robot can make adjustments based on real-time ground feedback, optimize the contact angle and posture. By measuring the touchdown pressure and slip adjustment value of the foot in real time, the robot can make dynamic adjustments according to the actual situation of each step, reduce slip caused by changes in ground friction or obstacles, thereby effectively improving the positioning accuracy and stability. By obtaining and analyzing multiple angular accelerations and linear accelerations in real time, combined with the local motion trajectories of the bionic robot in each preset section, more accurate motion state information can be provided for the robot. By obtaining three-dimensional environmental images in each local motion trajectory, corresponding local optimized maps can be formed, and the robot can flexibly adjust its own motion path and foot end contact angle according to these optimized maps. By obtaining three-dimensional environmental images at multiple consecutive moments and generating corresponding environmental maps based on these images, the changes in the surrounding environment can be dynamically updated, which enables the robot to make adjustments based on the latest environmental data, optimize the positioning and inspection paths. By integrating multiple total point cloud data sets, a high-precision three-dimensional positioning and inspection model can be obtained. These data not only help the bionic robot achieve efficient positioning, but also further provide high-quality map information, thereby optimizing the inspection path and improving the overall inspection efficiency and safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 It is a schematic flowchart of the method according to an embodiment of the present invention.

[0071] Figure 2 It is a schematic structural diagram of the system according to an embodiment of the present invention.

[0072] Figure 3 It is a schematic internal structure diagram of a computer device according to an embodiment of the present application.

[0073] The realization, functional features and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0074] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0075] As Figures 1 - 3 shown, the present application provides a method for inspecting and positioning an intelligent bionic robot at a grain terminal, including:

[0076] S1. Obtain the ground characteristics of multiple preset sections of the grain terminal and multiple initial foot end joint angles of the bionic robot, and adjust each initial foot end joint angle of the bionic robot according to each of the initial foot end joint angles and each ground characteristic;

[0077] S2. Obtain the position characteristics of the bionic robot after adjusting each initial foot joint angle and multiple foot touchdown pressure values, and obtain the corresponding slip adjustment value according to the position characteristics and the multiple foot touchdown pressure values;

[0078] S3. Adjust the foot contact points of the bionic robot according to each of the slip adjustment values, and obtain multiple angular accelerations and linear accelerations of the bionic robot in each preset section in real time during the adjustment process;

[0079] S4. Generate the local motion trajectory of the bionic robot in each preset section according to the multiple angular accelerations and linear accelerations, and obtain the three-dimensional environment images at multiple consecutive moments of the grain terminal in each local motion trajectory;

[0080] S5. Obtain the corresponding local environment map according to the multiple three-dimensional environment images, and obtain the corresponding local optimized map according to each local environment map;

[0081] S6. Obtain the corresponding total point cloud data set according to each local motion trajectory, and generate a three-dimensional positioning inspection model according to the multiple total point cloud data sets and the local optimized map, so that the bionic robot performs inspection positioning according to the three-dimensional positioning inspection model.

[0082] As described in the above steps S1 - S6, currently, when a bionic robot is performing inspection, it usually optimizes the trajectory by calculating the error between the current position and the target position. However, grain terminals usually have obstacles such as uneven ground, scattered items, and stacked goods. These factors cause the ground to be uneven, and there may be uneven friction on the surface. The existing bionic robot inspection and positioning technology lacks adjustment of the foot - end contact angle, which causes the robot to be unable to adapt to the ground changes in a timely manner when facing irregular terrain, resulting in positioning errors. Furthermore, it leads to the bionic robot missing some areas that need to be inspected, thus affecting the efficiency and quality of the inspection. The present invention adjusts each initial foot - end joint angle of the bionic robot based on the ground features of multiple preset sections of the grain terminal and the multiple initial foot - end joint angles of the bionic robot, and obtains the corresponding slip adjustment value through the position features of the bionic robot after each initial foot - end joint angle adjustment and multiple foot - bottom contact pressure values. The foot contact point is adjusted according to each slip adjustment value, where the foot contact point refers to the position where the foot of the bionic robot contacts the ground, the initial foot - end joint angle refers to the initial joint - angle configuration of the robot's foot - end (similar to the foot of a biological robot), that is, before the robot starts walking or working, the preset angles of each joint (such as toes, ankles, etc.) at the foot - end, and the foot - bottom contact pressure value refers to the magnitude of the pressure between the sole of the foot and the ground when each foot - end of the robot contacts the ground. The slip adjustment value refers to the amount of slip that needs to be adjusted due to factors such as uneven ground and changing friction during the contact between the robot's foot - end and the ground. Since grain terminals usually have obstacles such as uneven ground, scattered items, and stacked goods, resulting in uneven ground, these complex ground features may cause existing inspection robots to be unable to accurately track the target position and generate large errors.By dynamically adjusting each initial foot joint angle, the bionic robot can better adapt to these irregular terrains, thereby reducing the errors caused by terrain changes. By adjusting the foot end contact angle according to the ground characteristics and foot pressure values, the robot can make adjustments based on real-time ground feedback, optimize the contact angle and posture, and thus improve the stability and accuracy in complex environments. By measuring the ground contact pressure and slip adjustment value of the foot in real time, the robot can make dynamic adjustments according to the actual situation of each step. This mechanism enables the robot to adapt to different ground conditions, reduce slip caused by changes in ground friction or obstacles, and thus effectively improve the positioning accuracy and stability. By adjusting the foot end contact angle and combining the ground contact pressure and position characteristics for slip adjustment, it is possible to accurately capture the impact of ground changes on the robot, thereby improving the positioning accuracy. By adjusting according to the ground characteristics and foot end angles to adapt to environmental changes, rather than relying on excessive complex calculations, the control logic of the robot system can be simplified, while improving the system's response speed and adaptability. Through the adjustment of the foot end angle and pressure, the robot can better distribute the load, optimize the way the foot contacts the ground, reduce the risk of slip and tilt, and thus enhance the stability and passability of the robot. In the present invention, by obtaining in real time multiple angular accelerations and linear accelerations of the bionic robot in each preset section during the adjustment process, generating a local motion trajectory of the bionic robot in each preset section based on the multiple angular accelerations and linear accelerations, obtaining three-dimensional environmental images at multiple consecutive moments in each local motion trajectory of the grain terminal, obtaining a corresponding local environmental map through the multiple three-dimensional environmental images, obtaining a corresponding local optimized map according to each local environmental map, then obtaining a corresponding total point cloud data set through each of the local motion trajectories, and generating a three-dimensional positioning and inspection model based on the multiple total point cloud data sets and the local optimized maps, so that the bionic robot can perform inspection and positioning according to the three-dimensional positioning and inspection model. Among them, the point cloud data set is a set of spatial points collected by a lidar or other sensors, and each point contains three-dimensional coordinates, representing the surface of an object in the environment. By obtaining and analyzing multiple angular accelerations and linear accelerations in real time and combining the local motion trajectory of the bionic robot in each preset section, more accurate motion state information can be provided for the robot, which enables the robot to better adapt to different terrains and environmental changes, thereby reducing positioning errors caused by factors such as uneven ground or uneven friction. By obtaining three-dimensional environmental images in each local motion trajectory, a corresponding local optimized map can be formed, and the robot can flexibly adjust its own motion path and foot end contact angle according to these optimized maps, thereby effectively coping with irregular terrains, reducing errors and ensuring the completion of the inspection task. By obtaining three-dimensional environmental images at multiple consecutive moments and generating corresponding environmental maps based on these images, the changes in the surrounding environment can be dynamically updated.For example, factors such as cargo stacking and item movement can affect the ground condition. The real-time three-dimensional environmental image can reflect these changes, and then generate a more accurate local optimization map, which enables the robot to make adjustments based on the latest environmental data, optimize the positioning and inspection path. By integrating multiple total point cloud data sets, a high-precision three-dimensional positioning and inspection model can be obtained. These data not only help the bionic robot achieve efficient positioning, but also further provide high-quality map information, and then optimize the inspection path. In this way, the robot can more intelligently identify and avoid obstacles, thus avoiding collisions or stagnation, and improving the overall inspection efficiency and safety. By generating a three-dimensional positioning and inspection model according to the local optimization map, the robot can automatically optimize its own movement path during driving. The bionic robot can adjust the foot-end contact angle and other movement parameters according to real-time feedback to adapt to irregular terrains and complex environments, thereby improving the inspection efficiency and the adaptability of the robot in different terrains. Therefore, by comprehensively using angular acceleration, linear acceleration, three-dimensional environmental images, point cloud data, and local optimization maps, the problem of positioning errors of bionic robots in irregular terrains is effectively overcome, and the adaptability, accuracy, and autonomy of the robot in complex environments are improved, providing a more reliable and efficient solution for inspection tasks in complex environments such as grain terminals.

[0083] In one embodiment, step S1 of adjusting each initial foot-end joint angle of the bionic robot according to each of the initial foot-end joint angles and each ground feature includes:

[0084] S11. Obtain height data of multiple adjacent points according to each of the ground features, and obtain a corresponding ground normal vector according to the multiple height data;

[0085] S12. Obtain a corresponding ground horizontal slope and a ground vertical slope according to each of the ground normal vectors, and obtain a corresponding ground slope angle according to each of the ground horizontal slope and the ground vertical slope;

[0086] S13. Obtain a corresponding foot-end contact plane normal vector of the bionic robot according to each of the initial foot-end joint angles, and obtain a corresponding initial foot-end grounding angle according to each of the foot-end contact plane normal vectors and the ground normal vectors;

[0087] S14. Determine whether each of the initial foot-end grounding angles is different from the ground slope angle;

[0088] If the initial foot-end grounding angle is different from the ground slope angle, obtain a corresponding error angle according to each of the initial foot-end grounding angles and the ground slope angle;

[0089] S15. Adjust the initial foot-end joint angles of the bionic robot according to each of the error angles.

[0090] As described in the above steps S11 - S15, the slope angle refers to the inclination angle of the ground relative to the horizontal plane or the reference plane. The ground normal vector refers to the direction perpendicular to the ground, indicating the direction of the ground inclination. The foot - end contact plane normal vector refers to the normal - direction vector of the foot - contact plane when the robot's foot - end contacts the ground. The initial foot - end grounding angle refers to the initial angle of the robot's foot - end when it contacts the ground. In the present invention, the ground normal vector is obtained through the height data of multiple adjacent points of the ground features. The ground horizontal slope and the ground vertical slope are obtained based on the ground normal vector, and the ground slope angle is obtained based on the ground horizontal slope and the ground vertical slope. By obtaining the height data of multiple adjacent points and calculating the ground normal vector, the inclination and direction of the ground can be accurately described, which can help the robot precisely understand the slope characteristics of the ground. By adjusting the contact angle of the foot - end according to the ground slope angle, the robot can better adapt to complex terrains and avoid problems such as slippage and unstable walking. Bionic robots often face slippage problems caused by inaccurate foot - end contact angles in complex ground environments. By accurately calculating the slope and normal vector of the ground, precise feedback on the ground features can be provided to help adjust the contact angle of the robot's foot - end, thereby reducing the slippage phenomenon. By obtaining the ground slope angle and adjusting the foot - end contact angle accordingly, the robot can maintain a better contact in complex terrains and reduce the walking instability caused by uneven ground. Traditional positioning methods may not effectively combine ground features with visual data, resulting in positioning errors and unstable trajectories.By combining the calculation of the ground slope angle with visual positioning data, the robot can model the environment more accurately. Thus, in complex environments such as grain terminals, it can precisely navigate and position. The combined motion trajectory can ensure that the robot maintains a stable travel route in a dynamically changing environment. The calculation of the ground slope angle and normal vector provides the robot with more accurate environmental data, which can help optimize the positioning process during the dynamic update of the environmental map. By combining these ground features with visual positioning information, it can better reduce the trajectory calculation errors caused by complex environments or sensor errors in traditional methods. By introducing ground slope information into the global positioning process, the robot can more accurately identify and adapt to environmental changes under different ground conditions. This can reduce the trajectory drift or errors caused by uneven ground for the robot and improve the stability of the motion trajectory. Obtain the normal vector of the foot contact plane of the bionic robot according to the initial foot end joint angle, and obtain the initial foot end grounding angle according to the normal vector of the foot contact plane and the ground normal vector, and judge whether the initial foot end grounding angle is different from the ground slope angle. If the initial foot end grounding angle is different from the ground slope angle, obtain the error angle according to the initial foot end grounding angle and the ground slope angle, and use the PID control algorithm to adjust the initial joint angle of the bionic robot according to the error angle until the initial foot end grounding angle is the same as the ground slope angle. In different ground environments, especially in complex terrains, the foot end contact angle of the bionic robot may not be accurately adjusted, resulting in poor stability and insufficient adaptability of the robot. By obtaining the normal vector of the foot contact plane according to the initial foot end joint angle and then using the PID control algorithm to adjust the joint angle, it can ensure that the angle of the foot end contacting the ground exactly matches the ground slope angle, thereby improving the stability and adaptability of the robot in complex terrains. By adjusting the foot end grounding angle in real time to match the ground slope angle, the bionic robot can better adapt to various uneven ground surfaces, avoid problems such as slipping and instability, and increase its mobility in complex environments. Using the PID control algorithm to perform real-time feedback and adjustment on the grounding angle error can enable the robot to have self-adaptability in a dynamic environment.For example, during the walking process of the robot, if the ground changes, the PID algorithm can automatically adjust the grounding angle of the foot end, making the movement trajectory of the robot more stable, avoiding the error accumulation caused by the mismatch of the contact angle. The present invention can perform global positioning by combining visual positioning data with the movement trajectory of the robot. The visual system can help the robot detect the ground slope and obstacles, while the movement trajectory data helps analyze the actual movement state of the robot. By combining these data for dynamic adjustment, the positioning accuracy can be improved, the positioning error can be reduced, and the global movement control ability of the robot can be enhanced. By optimizing the grounding angle of the foot end, the movement trajectory of the robot in a complex environment will be more accurate, reducing the occurrence of errors. The stable contact angle reduces the possibility of the robot tilting or being unstable, thus ensuring that the posture of the robot is more stable during the walking process and effectively reducing the risk of falling or sliding.

[0091] In one embodiment, step S2 of obtaining the slip adjustment value according to the position feature and multiple foot bottom contact pressure values includes:

[0092] S21. Obtain the bionic robot's center of gravity position information and each foot end position information according to the position feature, and obtain the corresponding foot weight vector according to the center of gravity position information and each foot end position information;

[0093] S22. Obtain the sum of torque vectors according to the multiple foot bottom contact pressure values and the foot weight vector, and determine whether the sum of torque vectors is zero;

[0094] If the sum of torque vectors is zero, it is determined that the bionic robot does not generate slip;

[0095] If the sum of torque vectors is not zero, it is determined that the bionic robot generates slip, and obtain the reference force, friction coefficient, real-time foot end contact angle and motion characteristics of the bionic robot at this time, where the motion characteristics include real-time speed, real-time acceleration, total mass and gravitational acceleration;

[0096] S23. Obtain the weight of the bionic robot according to the total mass and gravitational acceleration, and obtain the vertical normal force according to the weight of the bionic robot and the real-time foot end contact angle;

[0097] S24. Obtain the inertial force according to the total mass and real-time acceleration, and obtain the horizontal normal force according to the inertial force and the real-time foot end contact angle;

[0098] S25. Obtain the total normal force according to the horizontal normal force and the vertical normal force, and obtain the frictional force according to the total normal force and the friction coefficient;

[0099] S26. Obtain the slip adjustment factor according to the reference force and the frictional force, and obtain the slip adjustment value according to the slip adjustment factor, real-time speed and real-time acceleration.

[0100] As described in the above steps S21 - S26, slippage generally refers to the relative movement between the feet of the robot and the ground, usually manifested as one or more supporting feet failing to fully contact the ground, or uneven distribution of contact forces resulting in insufficient ground friction, thereby causing slippage. The foot weight vector refers to the vector between the end of the robot's foot and the center of gravity. The present invention obtains the corresponding foot weight vector through the bionic robot's center of gravity position information and each foot end position information, obtains the sum of torque vectors based on multiple foot bottom contact pressure values and the foot weight vector, and determines whether the sum of torque vectors is zero. If the sum of torque vectors is not zero, it is determined that the bionic robot has slippage, and at this time, the reference force, friction coefficient, real-time foot end contact angle, and motion characteristics of the bionic robot are obtained, where the motion characteristics include real-time speed, real-time acceleration, total mass, and gravitational acceleration. During the inspection process, the bionic robot may slip due to uneven ground, changing friction, or the presence of obstacles. By obtaining the position information of the center of gravity and each foot end and the contact pressure in real time, combined with the method of calculating the sum of torque vectors, it is possible to accurately detect whether the robot has slippage. If the sum of torque vectors is not zero, the system can determine that slippage has occurred, and thus timely take adjustment measures to prevent the robot from further losing stability or positioning error. By dynamically adjusting the slippage value, the robot can better adapt to these irregular grounds and objects, enabling it to operate more stably on uneven ground. By making real-time slippage adjustments according to the pressure value of each foot end and the torque vector, the robot can adjust the contact angle and position of its foot end at each step, enabling the robot to quickly adapt during terrain changes and improve positioning accuracy. By monitoring the center of gravity, foot end position information, and pressure value of the bionic robot in real time, accurately judging and dealing with the slippage phenomenon, the adaptability, stability, and positioning accuracy of the robot are improved. In complex and irregular ground environments, the bionic robot can automatically adjust the foot end contact angle to avoid slippage, thereby realizing efficient and stable inspection tasks in complex environments such as grain terminals. The weight of the bionic robot is obtained through the total mass and gravitational acceleration, the vertical normal force is obtained according to the weight of the bionic robot and the real-time foot end contact angle, the inertial force is obtained according to the total mass and real-time acceleration, and the horizontal normal force is obtained according to the inertial force and the real-time foot end contact angle. The total normal force is obtained according to the horizontal normal force and the vertical normal force, and the friction force is obtained according to the total normal force and the friction coefficient. The slippage adjustment factor is obtained through the reference force and the friction force, and the slippage adjustment value is obtained according to the slippage adjustment factor, real-time speed, and real-time acceleration. By obtaining the foot end contact angle in real time, the robot can adjust its gait to adapt to uneven and undulating ground. By adjusting the contact angle in real time, it helps the robot timely respond to changes in irregular terrain and avoid imbalance or deviation from the target path caused by inability to adapt to ground changes. By calculating the inertial force and the normal force, the robot can more accurately perceive its contact state with the ground and further precisely control the motion direction and trajectory, which not only enhances the robot's adaptability to terrain changes,It can also reduce the positioning error caused by the inability to adjust the foot-end contact angle in a timely manner. By calculating the friction force and slip adjustment factor, the robot can make adaptive adjustments according to the unevenness of the ground friction, reduce the impact of slip on positioning, and can adjust the path in real time based on information such as the current position of the robot, the foot-end contact angle, acceleration, and speed. Thus, it can ensure that the robot can continuously optimize its traveling trajectory according to the real-time environmental changes. The robot can quickly respond to changes in ground conditions, adjust its walking mode, ensure accurate arrival at the target position, and at the same time avoid stagnation or deviation caused by non-adaptation to ground changes. By monitoring and adjusting the foot-end contact angle and friction force in real time, the robot can adapt to different ground conditions, thereby reducing the interference of the external environment on its movement. The above method dynamically adjusts the motion state of the robot by integrating multiple factors (gravitational acceleration, contact angle, inertial force, friction force, etc.), significantly improving the adaptability of the bionic robot on complex and uneven ground. By optimizing the foot-end contact angle and reducing the impact of friction force unevenness on positioning accuracy, the robot can execute inspection tasks more accurately and stably, avoid positioning errors caused by uneven ground or obstacles, and thus improve the inspection efficiency and work reliability.

[0101] In one embodiment, step S3 of generating a local motion trajectory according to multiple angular accelerations and linear accelerations includes:

[0102] S31. Obtain the sampling frequency and initial features of the bionic robot, where the initial features include initial position information, initial linear velocity, initial angular velocity, and initial angle;

[0103] S32. Obtain the time step according to the sampling frequency, and calculate the corresponding real-time updated linear velocity according to the time step, initial linear velocity, and each linear acceleration, where the calculation formula is:

[0104] ;

[0105] where S(XS) represents the real-time updated linear velocity, C(XS) represents the initial linear velocity, X(JS) represents the linear acceleration, and S(BC) represents the time step;

[0106] S33. Obtain the real-time updated position according to each real-time updated linear velocity, initial position, linear acceleration, and time step, where the calculation formula is:

[0107] ;

[0108] where S(XW) represents the real-time updated position, S(XS) represents the real-time updated linear velocity, C(XW) represents the initial position, and S(BC) represents the time step;

[0109] S34. Obtain the corresponding real-time updated angular velocity according to the time step, the initial angular velocity, and each angular acceleration;

[0110] S35. Obtain the real-time updated angle according to each of the real-time updated angular velocities, angular accelerations, initial angles, and time steps, wherein the calculation formula for the real-time updated angle is the same as the calculation formula for the real-time updated position;

[0111] S36. Establish a three-dimensional time-angle-position coordinate axis with time as the X-axis, angle as the Y-axis, and position as the Z-axis, and plot the initial position information and the initial angular velocity as the starting point on the three-dimensional time-angle-position coordinate axis;

[0112] S37. Plot each real-time updated angle and real-time updated position corresponding to each time step as connection points on the three-dimensional time-angle-position coordinate axis;

[0113] S38. Connect the starting point and multiple connection points in sequence through curves to obtain a local motion trajectory.

[0114] As described in the above steps S31 - S38, the present invention obtains the sampling frequency, initial position information, initial linear velocity, initial angular velocity, and initial angle of the bionic robot, obtains the time step according to the sampling frequency, obtains the corresponding real-time updated linear velocity according to the time step, initial linear velocity, and each linear acceleration, obtains the real-time updated position according to each real-time updated linear velocity, initial position, linear acceleration, and time step, obtains the corresponding real-time updated angular velocity according to the time step, initial angular velocity, and each angular acceleration, obtains the real-time updated angle according to each real-time updated angular velocity, angular acceleration, initial angle, and time step, takes time as the X-axis, angle as the Y-axis, and position as the Z-axis to establish a three-dimensional time-angle-position coordinate axis, plots the initial position information and initial angular velocity as the starting point on the three-dimensional time-angle-position coordinate axis, plots each real-time updated angle and real-time updated position corresponding to each time step as connection points on the three-dimensional time-angle-position coordinate axis, and sequentially connects the starting point and multiple connection points through curves to obtain a local motion trajectory. Usually, in a complex ground environment, the ground may be uneven, there may be slippage or inconsistent friction. By real-time updating the linear velocity, angular velocity, position, and angle of the robot, the motion state of the robot on the uneven ground can be more accurately reflected, thereby effectively predicting and adjusting the motion trajectory of the robot. By real-time updating the position and angle of the robot according to the sampling frequency, time step, initial velocity, and acceleration, not only can the motion state of the robot be effectively simulated, but also the accuracy of trajectory calculation can be improved. Traditional positioning methods usually rely on single sensor data, which may have errors or lack sufficient flexibility. By combining visual positioning data and real-time updated motion trajectories, the accuracy and robustness of positioning can be improved. The robot can calculate a more accurate global position based on visual information and motion trajectories, thereby avoiding error accumulation caused by a single data source. The motion and foot-end contact angle of the bionic robot will have a greater impact on the stability of the robot. By real-time updating the angle and position, this method can timely adjust the motion state of the robot, avoid slippage caused by inaccurate contact angles, and thus improve the adaptability and stability of the robot. By establishing a three-dimensional time-angle-position coordinate axis and dynamically plotting the motion trajectory, the robot can real-time adjust its motion strategy to cope with sudden ground changes or obstacles, which enables the robot to more flexibly cope with complex environments and reduce path deviations caused by environmental changes. Through fine real-time updates and calculations of time steps, path drift caused by initial position errors or sensor noise can be effectively reduced. During long-term operation, the robot can maintain a high positioning accuracy by continuously correcting the real-time data of the angle, velocity, and position, reduce error accumulation, and ensure a more stable and accurate motion trajectory. By realizing dynamic trajectory optimization and real-time feedback adjustment, the robot can autonomously cope with complex environments without external intervention.This ability to autonomously adjust the motion trajectory makes the robot more intelligent, enabling it to complete tasks without manual intervention and reducing the burden of manual operation.

[0115] In one embodiment, step S5 of obtaining a local environment map based on the multiple three-dimensional environment images and obtaining a locally optimized map based on the local environment map includes:

[0116] S51. Divide each of the three-dimensional environment images into multiple local windows, and obtain a plurality of pixel points based on each local window;

[0117] S52. Obtain the horizontal gradient and vertical gradient of each pixel point, and obtain the sum of the squares of the horizontal gradients based on the multiple horizontal gradients;

[0118] S53. Obtain the sum of the squares of the vertical gradients based on the multiple vertical gradients, and obtain the sum of the products of the gradients based on the multiple vertical gradients and horizontal gradients;

[0119] S54. Generate an autocorrelation matrix based on the sum of the products of the gradients, the sum of the squares of the horizontal gradients, and the sum of the squares of the vertical gradients, and obtain the determinant value and trace value based on the autocorrelation matrix;

[0120] S55. Obtain a response function value based on the determinant value and trace value, and determine whether the response function value is greater than a preset threshold;

[0121] If the response function value is greater than the preset threshold, determine that the pixel point is a corner point;

[0122] S56. Obtain corresponding matching points based on the multiple corner points, and use the SLAM algorithm to construct a local environment map based on the multiple matching points;

[0123] S57. Extract feature points corresponding to each corner point based on the local environment map, and obtain a reprojection error based on the multiple feature points and corner points;

[0124] S58. Optimize and adjust the feature points in the local environment map based on the reprojection error to obtain a locally optimized map.

[0125] As described in the above steps S51 - S58, the present invention divides each three - dimensional environmental image into multiple local windows, obtains multiple pixel points according to each local window, obtains the sum of squared horizontal gradients through the horizontal gradients of the multiple pixel points, obtains the sum of squared vertical gradients through the vertical gradients of the multiple pixel points, and obtains the sum of gradient products according to the vertical gradients and horizontal gradients of the multiple pixel points. A self - correlation matrix is generated through the sum of gradient products, the sum of squared horizontal gradients, and the sum of squared vertical gradients, and the determinant value and trace value are obtained according to the self - correlation matrix. The response function value is obtained according to the determinant value and trace value, and it is determined whether the response function value is greater than a preset threshold. If the response function value is greater than the preset threshold, then the pixel point is determined to be a corner point. Among them, a corner point refers to a point in the point cloud data where the change in the local area is the most obvious, usually representing the edge or feature point of an object, having a relatively high curvature. Bionic robots need to handle complex environments such as uneven ground, scattered objects, and obstacles during the inspection process. Traditional positioning methods may not be able to handle these irregular terrains well. By dividing local windows in each three - dimensional environmental image and calculating different gradients, it can help the robot more accurately perceive and adapt to terrain changes. For example, the sum of squared horizontal and vertical gradients and the sum of gradient products can be used to better capture the undulations of the terrain, thereby improving the adaptability to irregular terrain changes. By calculating the determinant value and trace value of the self - correlation matrix and judging corner points according to the response function value, the robot can quickly identify key terrain features in a dynamic environment. The detection of corner points can help the robot identify feature points on the ground (such as the edges and protrusions of obstacles), thereby optimizing path planning and obstacle avoidance strategies, and avoiding positioning deviations caused by uneven ground or obstacles. By detecting and calculating the gradients of local images, it can help the robot identify areas with uneven friction, thereby adjusting the foot - end contact angle and gait, improving the robot's motion adaptability in complex environments, and reducing sliding errors caused by uneven friction. Through the gradient calculation based on local windows and the determination of the response function, the bionic robot can continuously make real - time adjustments according to sensor data. This method enables the robot to be more self - adaptive when facing dynamic and complex terrains, respond promptly to ground changes, and improve the stability and reliability of the robot during the inspection process. Using gradient information and the response function, the robot can accurately identify and optimize the error between the current and target positions, reduce positioning errors caused by irregular terrains, and thus achieve precise trajectory optimization. By making real - time adjustments to the path and positioning, it can effectively avoid error accumulation caused by ground undulations or obstacles. Among them, the trace of the self - correlation matrix reflects the total sum of gradient changes in the horizontal and vertical directions of the local area of the image, and the determinant of the self - correlation matrix represents the deformation ability of the matrix, which is a measure of the intensity of gradient changes. The larger the determinant, the more complex the local change of the image in this area, which may correspond to corner points or edges. The response function is the key value for measuring whether each pixel point is a corner point. The response function formula is:, where R represents the response function, H(LS) represents the determinant value, and β represents an empirical constant, usually taking values between 0.04 and 0.Between 0 and 6, Z(JZ) represents the trace value. If the value of R is large, it usually indicates that the local area around this point has strong changes and has gradient information in multiple directions, which usually corresponds to the corner points of the image. The product sum of gradients reflects the mutual relationship of gradients in two directions of the image and is often called gradient covariance. The sum of the squares of the horizontal gradients and the sum of the squares of the vertical gradients respectively represent the change intensities of the image in the horizontal and vertical directions and are called the square of the gradient energy or gradient amplitude. In the present invention, corresponding matching points are obtained through multiple corner points, and the SLAM algorithm is used to construct a local environment map according to multiple matching points. Feature points corresponding to each corner point are extracted according to the local environment map, and the reprojection error is obtained according to multiple feature points and corner points. The feature points in the local environment map are optimized and adjusted according to the reprojection error to obtain a locally optimized map. Among them, the reprojection error refers to the error between the original point and the projected point when the three-dimensional point cloud data is projected onto a two-dimensional plane through the projection model of a camera or other sensors. Through corner point extraction and calculation of matching points, the bionic robot can identify key feature points in the environment. These feature points construct a local environment map through the SLAM algorithm, helping the robot identify and locate significant features (such as scattered items, stacked goods and other obstacles) in its surrounding environment. It can not only cope with complex ground conditions, but also improve the accuracy and reliability of the robot in a complex environment. By calculating the reprojection error according to the feature points and corner points and through optimization and adjustment, the position of the feature points in the local environment map can be accurately adjusted, thereby reducing the error caused by terrain changes. The optimized local map can better adapt to the irregular ground and improve the stability of positioning. Through the SLAM algorithm and corner point matching, the robot can obtain ground features in real time and dynamically adjust its path according to these features, so as to help the robot better adapt to ground undulations and obstacles and avoid positioning errors caused by ground irregularities. The combination of the SLAM algorithm with corner point optimization and reprojection error correction can help the robot more accurately perceive the ground conditions and, according to the optimization information of the local map, timely adjust the gait or the foot-end contact angle, thereby reducing the influence of uneven friction on the robot's movement. The SLAM algorithm enables the robot to update its local environment map in real time during the inspection process and adjust its path planning according to the new map. This dynamic update function enables the robot to timely respond to environmental changes, especially changes in ground friction and obstacles, and avoid error accumulation caused by static path planning in traditional methods. The SLAM and optimization algorithms, by obtaining environmental feature points and error feedback in real time, help the robot adjust the foot-end contact angle and gait according to the real-time ground conditions. Through corner point matching, local map optimization and correction of the reprojection error, the robot can perform error correction in real time and reduce error accumulation caused by irregular terrain or obstacles. This not only improves the accuracy of robot positioning, but also greatly improves the inspection efficiency.

[0126] In one embodiment, step S6 of obtaining a corresponding total point cloud data set according to each local motion trajectory and generating a three-dimensional positioning inspection model based on the multiple total point cloud data sets and the local optimized map includes:

[0127] S61. Real-time collect the point cloud data sets to be registered at multiple preset positions of the grain terminal according to each local motion trajectory, and select the point cloud data set to be registered at the central position as the target point cloud data set;

[0128] S62. For each point to be registered in each point cloud data set to be registered, obtain the nearest neighbor matching point from the target point cloud data set by using the Euclidean distance metric method;

[0129] S63. Obtain the distance difference according to each nearest neighbor matching point and the point to be registered, and adjust the position of each point to be registered in each point cloud data set to be registered according to the distance difference to obtain the corresponding registered point cloud data set;

[0130] S64. Convert each registered point cloud data set to the global coordinate system and delete the duplicate registered points to obtain the total point cloud data set;

[0131] S65. Use the PSR surface reconstruction algorithm to convert each total point cloud data set and the corresponding local optimized map into a local three-dimensional inspection model;

[0132] S66. Obtain the edge coincidence points of every two adjacent local three-dimensional inspection models, and combine and splice the multiple local three-dimensional inspection models according to each edge coincidence point to obtain the three-dimensional positioning inspection model.

[0133] As described in the above steps S61 - S66, the present invention collects the point cloud data sets to be registered at multiple preset positions of the grain terminal in real time through each local motion trajectory, and selects the point cloud data set at the central position as the target point cloud data set. For each point to be registered in each point cloud data set to be registered, the nearest neighbor matching point is obtained from the target point cloud data set by using the Euclidean distance metric method. The distance difference is obtained based on each nearest neighbor matching point and the point to be registered, and the position of each point to be registered in each point cloud data set to be registered is adjusted according to the distance difference to obtain the corresponding registered point cloud data set. Among them, the nearest neighbor matching point refers to the point with the smallest distance from each point in the point cloud data set to be registered found from the target point cloud data set through the Euclidean distance metric. By matching each local point cloud data set with the target point cloud data set, the position of the robot in the complex environment can be updated and adjusted in real time. This point cloud registration method helps to accurately identify the difference between the current position and the target position of the robot, thereby improving the positioning accuracy. On the irregular ground such as a grain terminal, the robot can quickly adapt to terrain changes through point cloud matching, avoiding the error accumulation that may occur in traditional methods. Using the Euclidean distance to calculate the nearest neighbor matching point can accurately capture the difference between the point cloud data set to be registered and the target point cloud. This accuracy helps the robot to reduce the error caused by mis - matching when dealing with obstacles such as scattered items and stacked goods, thus maintaining accurate positioning. Through the point cloud registration technology, the positions of these irregular terrains or obstacles can be identified in real time, and the path and posture of the robot can be adjusted, enabling more accurate identification of obstacles and avoiding positioning errors caused by uneven ground or obstacles during the inspection process. Through the positioning optimization based on point cloud registration, the robot can adjust the contact angle of its foot end in real time to adapt to the changes of different grounds. For example, when the robot passes through uneven ground, it can adjust the contact angle and position of the foot end according to the point cloud data to avoid slipping or unstable movement. This is especially important for bionic robots. Since the ground of the grain terminal may have uneven friction, traditional methods may be difficult to cope with the positioning errors caused by friction changes. Through real - time point cloud registration and distance difference adjustment, the robot can dynamically adjust the contact angle of the foot end, reducing the influence caused by uneven friction and improving the stability of the robot on irregular ground. Each time the point cloud data set is registered, the robot adjusts the point cloud according to the distance difference. This local optimization process helps the robot to continuously improve its understanding of the environment during the inspection process, thereby continuously enhancing its adaptive ability. Especially when facing complex terrains or environmental changes, it can make timely adjustments to reduce positioning errors. Through local point cloud optimization and error adjustment, the robot can generate or update the path planning in real time. By collecting and registering the point cloud data in real time through each local motion trajectory, the error can be adjusted and corrected in time, maintaining high stability and reliability. Through accurate point cloud registration and dynamic adaptation, the robot can complete the inspection task more smoothly.Especially in an environment with a high degree of uncertainty such as a grain terminal, it can ensure the efficient completion of tasks. By converting each registered point cloud data set into the global coordinate system and deleting duplicate registered points, the total point cloud data set is obtained. The PSR surface reconstruction algorithm is used to convert each total point cloud data set and the corresponding local optimization map into a local three-dimensional inspection model. By obtaining the edge coincidence points of every two adjacent local three-dimensional inspection models and combining and splicing multiple local three-dimensional inspection models according to each edge coincidence point, a three-dimensional positioning inspection model is obtained. By combining and splicing multiple local three-dimensional inspection models into a three-dimensional positioning inspection model, the topographic features of the inspection area can be presented more comprehensively. This helps the robot to perform precise positioning and navigation in a complex and uneven environment, especially in places like grain terminals where there are uneven ground and obstacles. The detailed topographic information provided by the three-dimensional inspection model enables the robot to better adapt to the changes in the ground, thereby reducing the positioning errors caused by terrain changes. Through the PSR surface reconstruction algorithm and the splicing of local three-dimensional models, the robot can perceive the subtle changes in the ground in real time and optimize its movement path according to this information. In this way, when the robot faces irregular terrain, it can more accurately estimate its position, thereby avoiding positioning errors caused by uneven friction and the influence of obstacles. Through the conversion and reconstruction of the total point cloud data, not only can the ground model be reconstructed more accurately, but also the dynamic adaptation between the robot's foot end and the ground can be realized by analyzing the relationship between the ground and the contact angle of the foot end. By combining high-precision point cloud data and three-dimensional modeling technology, the present invention can generate an accurate topographic model in real time in an uneven and complex environment, enabling the robot to autonomously adjust and optimize the path. This not only reduces the need for manual intervention, but also improves the inspection efficiency and accuracy of the bionic robot, and adapts to more complex working environments, such as places with sundries and stacked objects in grain terminals. By obtaining the edge coincidence points of two adjacent local three-dimensional inspection models and using these coincidence points for the splicing and combination of local models, a more complete and coherent three-dimensional positioning model can be realized. This method can effectively avoid the limitations of a single model. Especially when performing large-scale inspections, it can reduce errors caused by changes in perspective or local environment, thereby increasing the success rate of the entire inspection task.

[0134] The present application also provides an inspection and positioning system for an intelligent bionic robot in a grain terminal, including:

[0135] A first adjustment module, configured to obtain the ground features of multiple preset sections of the grain terminal and the initial foot end joint angles of the bionic robot, and adjust each initial foot end joint angle of the bionic robot according to each of the initial foot end joint angles and each ground feature;

[0136] The first acquisition module is used to acquire the position features of the bionic robot after adjusting each initial foot end joint angle and multiple foot contact bottom pressure values, and acquire the corresponding slip adjustment values according to the position features and multiple foot contact bottom pressure values;

[0137] The second acquisition module is used to adjust the foot contact points of the bionic robot according to each of the slip adjustment values, and acquire multiple angular accelerations and linear accelerations of the bionic robot within each preset section in real time during the adjustment process;

[0138] The first generation module is used to generate the local motion trajectory of the bionic robot within each preset section according to multiple angular accelerations and linear accelerations, and acquire three-dimensional environment images of the grain terminal at multiple consecutive moments within each local motion trajectory;

[0139] The third acquisition module is used to acquire the corresponding local environment map according to multiple of the three-dimensional environment images, and acquire the corresponding local optimized map according to each of the local environment maps;

[0140] The second generation module is used to acquire the corresponding total point cloud data set according to each of the local motion trajectories, and generate a three-dimensional positioning inspection model according to multiple of the total point cloud data sets and the local optimized map, so that the bionic robot performs inspection positioning according to the three-dimensional positioning inspection model.

[0141] In one embodiment, the first adjustment module includes:

[0142] The first acquisition unit is used to acquire the height data of multiple adjacent points according to each of the ground features, and acquire the corresponding ground normal vector according to the multiple height data;

[0143] The second acquisition unit is used to acquire the corresponding ground horizontal slope and ground vertical slope according to each of the ground normal vectors, and acquire the corresponding ground slope angle according to each of the ground horizontal slope and ground vertical slope;

[0144] The third acquisition unit is used to acquire the corresponding foot end contact plane normal vector of the bionic robot according to each of the initial foot end joint angles, and acquire the corresponding initial foot end grounding angle according to each of the foot end contact plane normal vectors and the ground normal vector;

[0145] The judgment unit is used to judge whether each of the initial foot end grounding angles is different from the ground slope angle;

[0146] If the initial foot end grounding angle is different from the ground slope angle, then acquire the corresponding error angle according to each of the initial foot end grounding angles and the ground slope angle;

[0147] The adjustment unit is used to adjust the initial foot end joint angle of the bionic robot according to each of the error angles.

[0148] It should be noted that each module and unit of the intelligent bionic robot in the grain terminal inspection and positioning system corresponds one by one to the steps in the method for inspecting and positioning the intelligent bionic robot in the grain terminal.

[0149] As Figure 3 shown, the present application also provides a computer device, which can be a server, and its internal structure can be as Figure 3 shown. The computer device includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer design is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store all the data required in the process of the method for inspecting and positioning the intelligent bionic robot in the grain terminal. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it realizes the method for inspecting and positioning the intelligent bionic robot in the grain terminal.

[0150] Those skilled in the art can understand that Figure 3 the structure shown in

[0151] is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied.

[0152] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium provided in this application and used in the embodiments can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or an external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0153] It should be noted that in this text, the terms "include", "comprise", or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that includes a series of elements includes not only those elements but also other elements not expressly listed, or elements that are inherent to such process, apparatus, article, or method. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, apparatus, article, or method that includes the element.

[0154] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structural or equivalent process transformation made by using the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A method for inspection and positioning of an intelligent bionic robot at a grain terminal, characterized in that: include: Acquire ground features of multiple preset sections of the grain terminal and multiple initial foot end joint angles of the bionic robot, and adjust each initial foot end joint angle of the bionic robot according to each of the initial foot end joint angles and each ground feature; Obtaining the position characteristics and multiple foot bottoming pressure values ​​of the bionic robot after each initial foot end joint angle adjustment, and obtaining corresponding slip adjustment values ​​according to the position characteristics and multiple foot bottoming pressure values; Adjusting the foot contact point of the bionic robot according to each of the slip adjustment values, and acquiring in real time multiple angular accelerations and linear accelerations of the bionic robot in each preset road section during the adjustment process; Generate a local motion trajectory of the bionic robot in each preset section according to multiple angular accelerations and linear accelerations, and obtain a three-dimensional environmental image of the grain terminal at multiple consecutive moments in each local motion trajectory; Dividing each of the three-dimensional environment images into a plurality of local windows, and acquiring a plurality of pixel points according to each of the local windows; Obtaining the horizontal gradient and the vertical gradient of each pixel point, and obtaining the sum of squares of the horizontal gradients according to the multiple horizontal gradients; Obtaining a vertical gradient square sum according to the plurality of vertical gradients, and obtaining a gradient product sum according to the plurality of vertical gradients and horizontal gradients; Generate an autocorrelation matrix according to the gradient product sum, the horizontal gradient square sum, and the vertical gradient square sum, and obtain a determinant value and a trace value according to the autocorrelation matrix; Obtaining a response function value according to the determinant value and the trace value, and determining whether the response function value is greater than a preset threshold; If the response function value is greater than a preset threshold, the pixel point is determined to be a corner point; Acquire corresponding matching points according to the plurality of corner points, and construct a local environment map according to the plurality of matching points using a SLAM algorithm; Extracting feature points corresponding to each corner point according to the local environment map, and obtaining a reprojection error according to a plurality of the feature points and corner points; Optimizing and adjusting the feature points in the local environment map according to the reprojection error to obtain a local optimized map; According to each of the local motion trajectories, the point cloud data sets to be registered at multiple preset positions of the grain terminal are collected in real time, and the point cloud data set to be registered at the central position is selected as the target point cloud data set; For each point to be registered in each of the point cloud data sets to be registered, the nearest neighbor matching point is obtained from the target point cloud data set by using the Euclidean distance measurement method; Obtaining a distance difference between each of the nearest neighbor matching points and the point to be registered, and adjusting the position of each of the point to be registered in each of the point cloud data sets to be registered according to the distance difference to obtain a corresponding registration point cloud data set; Each registered point cloud data set is converted into a global coordinate system and duplicate registration points are deleted to obtain a total point cloud data set; The PSR surface reconstruction algorithm is used to convert each total point cloud data set and the corresponding local optimization map into a local three-dimensional inspection model; The edge overlap points of each two adjacent local three-dimensional inspection models are obtained, and multiple local three-dimensional inspection models are combined and spliced ​​according to each of the edge overlap points to obtain a three-dimensional positioning inspection model, so that the bionic robot can perform inspection and positioning according to the three-dimensional positioning inspection model.

2. The method for inspection and positioning of an intelligent bionic robot at a grain terminal according to claim 1 is characterized in that: The step of adjusting each initial foot end joint angle of the bionic robot according to each initial foot end joint angle and each ground feature comprises: Acquire height data of a plurality of adjacent points according to each of the ground features, and acquire a corresponding ground normal vector according to the plurality of height data; Acquire a corresponding ground horizontal slope and a ground vertical slope according to each of the ground normal vectors, and acquire a corresponding ground slope angle according to each of the ground horizontal slope and the ground vertical slope; Acquire a foot contact plane normal vector corresponding to the bionic robot according to each of the initial foot joint angles, and acquire a corresponding initial foot contact angle according to each of the foot contact plane normal vectors and the ground normal vector; Determining whether each of the initial foot-end contact angles is different from a ground slope angle; If the initial foot-end contact angle is different from the ground slope angle, obtaining a corresponding error angle according to each of the initial foot-end contact angle and the ground slope angle; The initial foot end joint angle of the bionic robot is adjusted according to each of the error angles.

3. The method for inspection and positioning of an intelligent bionic robot at a grain terminal according to claim 1 is characterized in that: The step of obtaining the slip adjustment value according to the position feature and a plurality of foot bottom contact pressure values ​​comprises: Acquire the center of gravity position information and each foot end position information of the bionic robot according to the position feature, and acquire the corresponding foot weight vector according to the center of gravity position information and each foot end position information; Obtaining a torque vector sum according to the plurality of foot bottom contact pressure values ​​and foot weight vectors, and determining whether the torque vector sum is zero; If the sum of the torque vectors is zero, it is determined that the bionic robot has not slipped; If the moment vector sum is not zero, it is determined that the bionic robot has slipped, and the reference force, friction coefficient, real-time foot end contact angle and motion characteristics of the bionic robot at this time are obtained, wherein the motion characteristics include real-time speed, real-time acceleration, total mass and gravitational acceleration; Acquire the weight of the bionic robot according to the total mass and the gravitational acceleration, and acquire the vertical normal force according to the weight of the bionic robot and the real-time foot end contact angle; Acquire the inertial force according to the total mass and the real-time acceleration, and acquire the horizontal normal force according to the inertial force and the real-time foot end contact angle; Obtaining a total normal force according to the horizontal normal force and the vertical normal force, and obtaining a friction force according to the total normal force and a friction coefficient; A slip adjustment factor is obtained according to the reference force and the friction force, and a slip adjustment value is obtained according to the slip adjustment factor, the real-time speed and the real-time acceleration.

4. The method for inspection and positioning of an intelligent bionic robot at a grain terminal according to claim 1 is characterized in that: The step of generating a local motion trajectory according to a plurality of angular accelerations and linear accelerations comprises: Acquiring a sampling frequency and initial features of the bionic robot, wherein the initial features include initial position information, initial linear velocity, initial angular velocity, and initial angle; Acquire a time step according to the sampling frequency, and acquire a corresponding real-time updated linear velocity according to the time step, the initial linear velocity and each linear acceleration; Obtaining a real-time updated position according to each of the real-time updated linear velocities, initial positions, linear accelerations and time steps; Obtaining a corresponding real-time updated angular velocity according to the time step, the initial angular velocity and each angular acceleration; Acquire a real-time updated angle according to each of the real-time updated angular velocity, angular acceleration, initial angle and time step; With time as the X-axis, angle as the Y-axis, and position as the Z-axis, a three-dimensional coordinate axis of time-angle-position is established, and the initial position information and initial angular velocity are plotted on the three-dimensional coordinate axis of time-angle-position as the starting point; The real-time update angle and real-time update position corresponding to each time step are plotted as connection points on the three-dimensional coordinate axis of time-angle-position; The starting point and multiple connection points are sequentially connected through curves to obtain a local motion trajectory.

5. An intelligent bionic robot inspection and positioning system at a grain terminal, characterized in that: include: The first adjustment module is used to obtain ground features of multiple preset sections of the grain terminal and multiple initial foot end joint angles of the bionic robot, and adjust each initial foot end joint angle of the bionic robot according to each of the initial foot end joint angles and each ground feature; A first acquisition module is used to acquire the position characteristics and multiple foot bottoming pressure values ​​of the bionic robot after each initial foot end joint angle adjustment, and acquire corresponding slip adjustment values ​​according to the position characteristics and multiple foot bottoming pressure values; A second acquisition module is used to adjust the foot contact point of the bionic robot according to each of the slip adjustment values, and to acquire in real time multiple angular accelerations and linear accelerations of the bionic robot in each preset road section during the adjustment process; A first generating module is used to generate a local motion trajectory of the bionic robot in each preset section according to a plurality of angular accelerations and linear accelerations, and obtain a three-dimensional environmental image of the grain terminal at a plurality of consecutive moments in each local motion trajectory; A third acquisition module, used for dividing each of the three-dimensional environment images into a plurality of local windows, and acquiring a plurality of pixel points according to each of the local windows; Obtaining the horizontal gradient and the vertical gradient of each pixel point, and obtaining the sum of squares of the horizontal gradients according to the multiple horizontal gradients; Obtaining a vertical gradient square sum according to the plurality of vertical gradients, and obtaining a gradient product sum according to the plurality of vertical gradients and horizontal gradients; Generate an autocorrelation matrix according to the gradient product sum, the horizontal gradient square sum, and the vertical gradient square sum, and obtain a determinant value and a trace value according to the autocorrelation matrix; Obtaining a response function value according to the determinant value and the trace value, and determining whether the response function value is greater than a preset threshold; If the response function value is greater than a preset threshold, the pixel point is determined to be a corner point; Acquire corresponding matching points according to the plurality of corner points, and construct a local environment map according to the plurality of matching points using a SLAM algorithm; Extracting feature points corresponding to each corner point according to the local environment map, and obtaining a reprojection error according to a plurality of the feature points and corner points; Optimizing and adjusting the feature points in the local environment map according to the reprojection error to obtain a local optimized map; The second generation module is used to collect the point cloud data sets to be registered at multiple preset positions of the grain terminal in real time according to each of the local motion trajectories, and select the point cloud data set to be registered at the central position as the target point cloud data set; For each point to be registered in each of the point cloud data sets to be registered, the nearest neighbor matching point is obtained from the target point cloud data set by using the Euclidean distance measurement method; Obtaining a distance difference between each of the nearest neighbor matching points and the point to be registered, and adjusting the position of each of the point to be registered in each of the point cloud data sets to be registered according to the distance difference to obtain a corresponding registration point cloud data set; Each registered point cloud data set is converted into a global coordinate system and duplicate registration points are deleted to obtain a total point cloud data set; The PSR surface reconstruction algorithm is used to convert each total point cloud data set and the corresponding local optimization map into a local three-dimensional inspection model; The edge overlap points of each two adjacent local three-dimensional inspection models are obtained, and multiple local three-dimensional inspection models are combined and spliced ​​according to each of the edge overlap points to obtain a three-dimensional positioning inspection model, so that the bionic robot can perform inspection and positioning according to the three-dimensional positioning inspection model.

6. The intelligent bionic robot inspection and positioning system at a grain terminal according to claim 5 is characterized in that: The first adjustment module includes: A first acquisition unit, configured to acquire height data of a plurality of adjacent points according to each of the ground features, and acquire a corresponding ground normal vector according to the plurality of height data; A second acquisition unit, configured to acquire a corresponding ground horizontal slope and a ground vertical slope according to each of the ground normal vectors, and to acquire a corresponding ground slope angle according to each of the ground horizontal slope and the ground vertical slope; A third acquisition unit is used to acquire a foot end contact plane normal vector corresponding to the bionic robot according to each of the initial foot end joint angles, and to acquire a corresponding initial foot end ground contact angle according to each of the foot end contact plane normal vectors and the ground normal vector; A judging unit, used for judging whether each of the initial foot end contact angles is different from a ground slope angle; If the initial foot-end contact angle is different from the ground slope angle, obtaining a corresponding error angle according to each of the initial foot-end contact angle and the ground slope angle; The adjusting unit is used to adjust the initial foot end joint angle of the bionic robot according to each of the error angles.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images