A regional inspection system and method based on a quadruped bionic robot

Abstract

The application discloses a kind of based on four-foot bionic robot area inspection system and method, it is related to bionic robot security inspection technical field, the method includes the following steps: S100, through multimodal perception unit synchronous acquisition environmental data;S200, based on the terrain feature vector calculation terrain risk coefficient, and according to the value of terrain risk coefficient trigger ladder antiskid control chain;S300, based on visual barrier distance and sound wave moving speed calculation pass index;S400, through closed loop verification unit monitor S200's attitude stability and S300's path deviation's execution effect, when attitude stability or path deviation exceeds tolerance, trigger bidirectional optimization, reverse start S100's data reacquisition process, synchronous adjustment S200's risk coefficient and S300's pass index's calculation weight;S500, the optimization result of S400 is written into twin map model, generates and includes equipment displacement alarm and joint life prediction's inspection report, completes from perception to feedback's whole chain closed loop.

Description

Technical Field

[0001] This invention relates to the field of bionic robot security inspection technology, and in particular to a regional inspection system and method based on a quadrupedal bionic robot. Background Technology

[0002] With the continuous development of industrial automation and intelligent technologies, inspection robots, as a new type of intelligent equipment, have been widely used in various industrial scenarios, such as the safety monitoring and maintenance of important facilities in power systems, petrochemicals, airports, and ports. Inspection robots replace manual inspections through automation, enabling them to perform continuous and high-precision inspection tasks in complex and hazardous environments, thereby ensuring the safe operation of equipment and the environment, and reducing labor costs and safety risks.

[0003] Chinese patent application number 2025102583243 discloses a path planning method and system for an intelligent inspection bionic robot. It obtains a third travel delay time based on static obstacle information, and finally obtains an optimal path based on the second and third travel delay times. This can comprehensively evaluate the time cost of different path branches when encountering various obstacles, select the path with the minimum total delay time as the optimal path, and thus ensure that the intelligent inspection bionic robot completes the task in a shorter time or closer to the preset inspection time completion plan.

[0004] Based on the aforementioned existing technologies, the challenges of a quadrupedal bionic robot's area inspection system lie in the fact that existing SLAM technologies, such as laser SLAM and visual SLAM, are primarily designed for single-layer planar environments. In multi-story buildings, sensor data continuity is interrupted, and the robot cannot maintain cross-floor positioning accuracy. The quadrupedal robot needs to dynamically calculate the optimal cross-floor nodes and coordinate mechanical actions, such as adjusting its climbing posture. Furthermore, in low-light or textured environments, such as the staircases of old buildings, the quadrupedal robot's vision module cannot accurately extract stair features, causing the quadrupedal robot's stride pattern receiving and judgment module to mistakenly trigger the flat ground mode instead of the staircase mode, resulting in gait instability, falls, or task interruption. This leads to insufficient environmental adaptability. At the same time, a single vision module is susceptible to interference from changes in lighting, missing textures, or dynamic obstacles, resulting in feature point matching failures. It cannot accurately identify special surfaces such as oil stains and ice, and incorrectly adjusts the quadrupedal robot's gait, easily leading to slippage and loss of control. Moreover, it cannot accurately predict suddenly appearing dynamic obstacles such as pedestrians and vehicles, resulting in low obstacle avoidance path generation efficiency and gait control instability, posing a risk of falls.

[0005] Given the aforementioned challenges, effectively addressing the issues of gait adjustment for quadrupedal bionic robots on complex terrains, as well as the instability caused by climbing stairs and dynamic obstacle avoidance failures in multi-story buildings, has become a pressing problem for those skilled in the art.

[0006] Therefore, it is necessary to invent a regional inspection system and method based on a quadrupedal bionic robot to solve the above problems. Summary of the Invention

[0007] The purpose of this invention is to provide a regional inspection system and method based on a quadrupedal bionic robot to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a method for area inspection based on a quadrupedal bionic robot, comprising the following steps: S100: Environmental data is collected synchronously through a multimodal sensing unit. The depth information obtained by the acoustic sensor array and the surface hardness data obtained by the tactile sensor array are fused through a spatiotemporal alignment mechanism to generate a terrain feature vector. S200. Calculate the terrain risk coefficient based on the terrain feature vector, and trigger the stepped anti-slip control chain according to the value of the terrain risk coefficient. S300: The passage index is calculated based on the distance to visual obstacles and the speed of sound wave movement, and the calculation result of the passage index is directly related to the obstacle avoidance decision chain. S400 monitors the attitude stability of S200 and the path deviation of S300 through the closed-loop verification unit. When the attitude stability or path deviation exceeds the tolerance, bidirectional optimization is triggered, and the data re-acquisition process of S100 is restarted in reverse. The calculation weights of the risk coefficient of S200 and the passage index of S300 are adjusted synchronously. The S500 incorporates the optimization results of the S400 into the twin map model, generating an inspection report that includes equipment displacement alarms and joint life predictions, thus completing a closed loop from perception to feedback.

[0009] Preferably, in S100, the spatiotemporal alignment mechanism includes binding the acoustic depth data, tactile hardness data and visual image to the same coordinate system through the spatiotemporal alignment module. Specifically, the acoustic array emits pulse signals at fixed time intervals to generate a depth point cloud; the tactile sensor collects the foot pressure distribution during the pulse interval; when the spatial deviation between the point cloud and the pressure distribution exceeds a set value, the visual sensor is triggered to supplement semantic information for calibration.

[0010] Preferably, in S200, the terrain risk coefficient is calculated by weighting the ratio of acoustic detection depth to tactile surface hardness, wherein the activation conditions and execution effects of the anti-slip control chain form a closed loop: the foot adhesion force increases stepwise with the increase of the terrain risk coefficient; the stride compression rate and the tactile friction coefficient are proportionally related. When the terrain risk coefficient exceeds the first threshold, the joint verification mechanism of sound wave and tactile sensation is activated to recalibrate the depth data. If the terrain risk factor still exceeds the first threshold after calibration, execute the stride compression command and activate the foot adsorption device.

[0011] Preferably, in S300, the passage index is jointly determined by the sound wave velocity, the tactile friction coefficient, and the visual obstacle distance. When the passage index is lower than a second threshold, a lateral avoidance gait is triggered and a sound wave avoidance signal is emitted. The obstacle avoidance decision chain includes a two-layer feedback mechanism: First layer: Calculate the lateral movement distance based on the traffic index; The second layer: dynamically adjusts the weighting of the passage index calculation based on the actual obstacle avoidance success rate.

[0012] Preferably, S400 includes the following steps: S410: Real-time monitoring of the quadruped robot's posture stability indicators through a closed-loop verification engine. S420. When the posture stability index is lower than the third threshold, increase the calculation weight of the tactile sensor data in the terrain feature vector, and simultaneously monitor the path tracking deviation data of the quadruped robot. S430. When the path tracking deviation exceeds the fourth threshold, reduce the calculation weight of visual sensor distance data in the passage index. S440, The optimized weight parameters are written into the terrain risk coefficient calculation rules of S200 in real time and S300 traffic index calculation rules; S450: Generate new motion control commands based on the updated calculation rules.

[0013] Preferably, in step S500, the remaining service life is predicted by real-time acquisition of the joint vibration spectrum of the robot. When the remaining service life is less than the preset time, the location of high-risk equipment is marked in the digital twin map, and the inspection path is optimized to prioritize the coverage of maintenance stations. At the same time, a maintenance alarm is generated and associated with the equipment displacement data.

[0014] Preferably, the normalization processing of the sound wave velocity and the tactile friction coefficient adopts a dynamic standardization method, which standardizes the original sound wave velocity and the tactile friction coefficient based on the statistical characteristics of historical data; the size of the data window used for calculating statistical features is dynamically adjusted according to the terrain risk coefficient. When the terrain risk coefficient is higher than the preset value, a smaller data window is used to improve real-time performance; otherwise, a larger data window is used to reduce the impact of noise; the standardized traffic index is obtained by weighted calculation of the standardized sound wave velocity value and the standardized tactile friction coefficient value, wherein the weight coefficient is dynamically allocated by the bidirectional optimization module.

[0015] Preferably, the compatibility between the reacquired data and the original data is ensured by a spatiotemporal alignment engine, which injects a global clock signal into the reacquired data to achieve time synchronization; the reacquired data is spatially aligned with the original point cloud coordinate system through visual semantic tags; data conflict resolution rules are established, and when the difference in terrain risk coefficient is greater than a preset value, the twin map is updated based on the reacquired data; when sensor data is not detected in the reacquired data, historical data is used and the confidence level is marked.

[0016] The present invention also provides a region inspection system for a quadrupedal bionic robot. The system is used to implement the above-mentioned region inspection method for a quadrupedal bionic robot, enabling the quadrupedal bionic robot to complete region inspection operations. The system includes: a multimodal perception unit, an intelligent decision-making unit, a closed-loop verification unit, and a twin map management unit.

[0017] Preferably, the multimodal sensing unit includes an integrated acoustic and tactile module, which is designed with a piezoelectric ceramic emitter and a PVDF tactile sensor on the same substrate, and outputs a data stream bound to the same coordinate system; the visual compensation module is activated when the deviation between the acoustic and tactile data is greater than a threshold. The intelligent decision-making unit includes a risk control chain module, which receives terrain feature vectors and outputs gait commands linked to terrain risk coefficients; and an obstacle avoidance decision chain module, which receives passage indexes and outputs control signals linked to avoidance distances.

[0018] The technical effects and advantages of this invention are as follows: This invention constructs a triple verification mechanism of sound waves, touch, and infrared, and integrates a multimodal physical characteristic fusion perception system. It employs sound pulse detection and tactile edge recognition to collaboratively verify the staircase features, and combines thermal imaging temperature difference to lock the staircase mode, eliminating gait instability caused by visual misjudgment. Simultaneously, through the fusion positioning technology of sound wave SLAM and tactile anchor points, based on real-time monitoring of tactile friction coefficient and sound wave reflectivity analysis, it dynamically triggers the increase of foot adhesion force to ensure the passability of special surfaces. Finally, it forms a closed-loop system of perception-localization-decision-execution, effectively solving the problem of gait instability in quadrupedal bionic robots caused by staircase misrecognition, terrain instability, and obstacle avoidance failure. Attached Figure Description

[0019] Figure 1 This is a schematic diagram illustrating the steps of the method of the present invention.

[0020] Figure 2 This is a schematic diagram of the gait decision-making process of the present invention.

[0021] Figure 3 This is a schematic diagram illustrating the gait feedback verification of the present invention.

[0022] Figure 4 This is a block diagram of the functional modules of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] First Embodiment This invention provides, for example Figures 1 to 4 The system shown is a regional inspection system based on a quadrupedal bionic robot. The system includes: a multimodal perception unit, an intelligent decision-making unit, a closed-loop verification unit, and a twin map management unit.

[0025] In this embodiment, the multimodal sensing unit includes an integrated acoustic and tactile module, which uses a piezoelectric ceramic emitter and a PVDF tactile sensor on the same substrate to output a data stream bound to the same coordinate system; the visual compensation module is activated when the deviation between the acoustic and tactile data is greater than a threshold. In this embodiment, the intelligent decision-making unit includes a risk control chain module for receiving terrain feature vectors and outputting gait commands linked to the terrain risk coefficient; and an obstacle avoidance decision chain module for receiving a passage index and outputting control signals linked to the avoidance distance.

[0026] In this embodiment, the closed-loop verification unit includes a data acquisition module and a parameter optimization module. The data acquisition module is used to monitor attitude stability and path deviation in real time, and the parameter optimization module is used to adjust the sampling weight of the sensing unit in reverse when the monitored value exceeds the limit. In this embodiment, the digital twin map management unit includes a dynamic labeling module and a predictive maintenance interface. The dynamic labeling module is used to trigger a map refresh when the device displacement exceeds a set value, and the predictive maintenance interface is used to receive joint life predictions and generate path optimization instructions.

[0027] It should be noted that the system also includes three channels for cross-module data interlocking. The first channel is for the optimization instructions of the closed-loop verification unit to be directly connected to the parameter register of the sensing unit; the second channel is for the control signals of the decision unit and the torque output of the execution unit to form closed-loop control; and the third channel is for the displacement alarm of the twin map to trigger the path replanning of the decision unit.

[0028] During use, an array of acoustic sensors is deployed in a ring along the boundary of the inspection area with a spacing of ≤2m. The acoustic reference reflection time is calibrated, and the depth error threshold is set to ±1cm. A tactile sensor array is embedded in the robot's feet and deployed in four zones according to the forefeet and hindfeet. The focal length and white balance of the vision sensor are calibrated, and the static obstacle recognition threshold is set to 0.3m. When the deviation between acoustic and tactile data is >5cm, the vision compensation module is automatically activated, and 70% of the computing resources are allocated for data fusion.

[0029] The risk control chain module loads terrain feature vector processing rules: the first threshold of the terrain risk coefficient is set to 0.6, and when the threshold is exceeded, the stride compression rate = tactile friction coefficient × 0.5; the foot adhesion dynamic adjustment rule: for every 0.1 increase in the risk coefficient, the adhesion force increases by 25N.

[0030] The second threshold is set to 0.3. When the threshold is below the threshold, the lateral avoidance distance is 1.5 × (1 - traffic index) meters. The weight of historical obstacle avoidance success rate is initialized to 0.6, and the weight of leakage risk penalty is 0.4. The attitude stability monitor sets the third threshold to 0.85 and the path deviation sets the fourth threshold to 0.3m. When attitude stability < 0.85: the weight of tactile data is increased to 0.7; a terrain feature vector correction instruction is generated. When path deviation > 0.3m: the weight of visual distance is reduced to 0.3, and the original data is automatically compressed into a motion trajectory matrix.

[0031] The dynamic labeling module sets the equipment displacement threshold to 10cm. When the displacement exceeds the threshold, the map is refreshed. The high-risk equipment marking rule is: when the joint life is <48h, it is marked as a red level three alarm.

[0032] It should be noted that this multimodal fusion inspection system innovatively solves the core operational defects of quadruped robots in complex environments: by fusing sound, touch, and vision into a three-dimensional perception system, it overcomes the problem of stride loss of control caused by the failure of step recognition; based on a closed-loop verification engine and cross-layer collaborative mechanism, it eliminates the risk of communication interruption relying on historical trajectories; by adopting electromagnetic shielding encapsulation and infrared dynamic compensation, it eradicates false alarms of sensors in extreme environments; by relying on a passage index model and sound source localization array, it resolves path conflicts caused by dynamic obstacle prediction bias; and by using a dynamic labeling module and collaborative path compression, it optimizes cross-layer inspection efficiency, ultimately forming a closed loop of perception-decision-execution-verification.

[0033] Second Embodiment Based on the system framework constructed above, the specific steps of the area inspection method based on a quadrupedal bionic robot proposed in this invention are described in detail in a process-oriented manner.

[0034] In this embodiment, the method includes the following steps: S100: Environmental data is collected synchronously through a multimodal sensing unit. The depth information obtained by the acoustic sensor array and the surface hardness data obtained by the tactile sensor array are fused through a spatiotemporal alignment mechanism to generate a terrain feature vector.

[0035] S200: Calculate the terrain risk coefficient based on the terrain feature vector, and trigger the stepped anti-slip control chain according to the value of the terrain risk coefficient.

[0036] It should be noted that the terrain risk coefficient is R. When the terrain risk coefficient R is greater than 0.6, the sound wave-tactile joint verification mechanism is activated to recalibrate the depth data. When the terrain risk coefficient R is still greater than 0.6 after calibration, the stride compression command (compression rate 40%~60%) is executed and foot adsorption is initiated.

[0037] S300: The passage index is calculated based on the visual obstacle distance and the sound wave speed, and the calculation result of the passage index is directly related to the obstacle avoidance decision chain.

[0038] It should be noted that the distance to the visual obstacle is... The speed of sound wave movement is The passage index is CPI. If CPI < 0.3, a crab-like gait is triggered for lateral avoidance and an acoustic avoidance signal is emitted. At the same time, the weight coefficient of the CPI formula is updated based on the avoidance success rate.

[0039] S400 monitors the attitude stability of S200 and the path deviation of S300 through a closed-loop verification unit. When the attitude stability or path deviation exceeds the tolerance, bidirectional optimization is triggered, and the data re-acquisition process of S100 is restarted in reverse. The calculation weights of the risk coefficient of S200 and the passage index of S300 are adjusted synchronously.

[0040] It should be noted that the attitude stability is S and the path deviation is D. When S < 0.85 or D > 0.3m, the terrain feature vector in S200 is re-acquired in reverse, and the obstacle avoidance decision parameters of S300 are optimized at the same time.

[0041] The S500 incorporates the optimization results of the S400 into the twin map model, generating an inspection report that includes equipment displacement alarms and joint life predictions, thus completing a closed loop from perception to feedback.

[0042] In S100 of this embodiment, the spatiotemporal alignment mechanism includes binding the acoustic depth data, tactile hardness data and visual image to the same coordinate system through the spatiotemporal alignment module. Specifically, the acoustic array emits pulse signals at fixed time intervals to generate a depth point cloud; the tactile sensor collects the foot pressure distribution during the pulse gap; when the spatial deviation between the point cloud and the pressure distribution exceeds a set value, the visual sensor is triggered to supplement semantic information for calibration.

[0043] It should be noted that the acoustic array can emit 40kHz pulses at 20ms intervals to generate depth point clouds; when the spatial deviation between the point cloud and the pressure distribution is >5cm, the visual sensor is triggered to supplement semantic information for calibration.

[0044] In S200 of this embodiment, the terrain risk coefficient is calculated by weighted calculation of the ratio of acoustic detection depth and the ratio of tactile surface hardness. The activation conditions and execution effects of the anti-slip control chain form a closed loop: the foot adhesion force increases stepwise with the increase of the terrain risk coefficient; the stride compression rate and the tactile friction coefficient are proportionally related.

[0045] When the terrain risk coefficient exceeds the first threshold, the sound wave and tactile joint verification mechanism is activated to recalibrate the depth data.

[0046] If the terrain risk factor still exceeds the first threshold after calibration, execute the stride compression command and activate the foot adsorption device.

[0047] It should be noted that the terrain risk factor R is calculated as follows: Where R is the terrain risk coefficient. For the measured depth of the acoustic step, For the maximum allowable step depth, For tactile surface hardness, The maximum safe surface hardness is represented by 0.7 for acoustic weighting and 0.3 for tactile weighting.

[0048] The activation conditions and execution effect of the anti-slip control chain form a closed loop. When R>0.6, the foot adhesion force F increases with the increase of R value: F=100+50×(R−0.6)N; Stride compression ratio η and tactile friction coefficient Interlock: η = 0.5 × .

[0049] In this implementation of S300, the passage index is jointly determined by the sound wave velocity, the tactile friction coefficient, and the visual obstacle distance. When the passage index is lower than a second threshold, a lateral avoidance gait is triggered and a sound wave avoidance signal is emitted. The obstacle avoidance decision chain includes a two-layer feedback mechanism: First layer: Calculate the lateral movement distance based on the passage index, that is, the lateral movement distance of the crab gait L = 0.5 × (1 − CPI) m.

[0050] The second layer: dynamically adjusts the weighting of the passage index calculation based on the actual obstacle avoidance success rate, that is: It should be noted that the calculation of the CPI (Consumer Price Index) is as follows: Among them, CPI is the traffic index, which is a quantitative indicator of the ability to pass safely in a dynamic environment. The lower the value, the higher the risk. The speed at which the sound source moves. Distance to visual obstacles.

[0051] In this embodiment, S400 includes the following steps: S410: Real-time monitoring of the quadruped robot's posture stability indicators through a closed-loop verification engine. S420. When the posture stability index is lower than the third threshold, increase the calculation weight of the tactile sensor data in the terrain feature vector, and simultaneously monitor the path tracking deviation data of the quadruped robot. S430. When the path tracking deviation exceeds the fourth threshold, reduce the calculation weight of visual sensor distance data in the passage index. S440. Write the optimized weight parameters into the terrain risk coefficient calculation rules of S200 and the traffic index calculation rules of S300 in real time. S450: Generate new motion control commands based on the updated calculation rules.

[0052] It should be noted that when the attitude stability S < 0.85, the weight of tactile data in the terrain feature vector is increased to 0.5; when the path deviation D > 0.3m, the visual distance in the accessibility index CPI is reduced. The weight is set to 0.3.

[0053] In step S500 of this implementation, the remaining service life of the robot is predicted by real-time acquisition of the joint vibration spectrum. When the remaining service life is less than the preset time, the location of high-risk equipment is marked in the digital twin map, and the inspection path is optimized to prioritize the coverage of maintenance stations. At the same time, a maintenance alarm is generated and associated with the equipment displacement data.

[0054] It should be noted that when the predicted remaining lifespan is less than 48 hours, the location of high-risk equipment is marked in the digital twin map, and the inspection route is optimized to prioritize the coverage of maintenance sites. At the same time, maintenance alarms are generated and associated with equipment displacement data.

[0055] In this implementation, the normalization of sound wave velocity and tactile friction coefficient adopts a dynamic standardization method. Based on the statistical characteristics of historical data, the original sound wave velocity and tactile friction coefficient are standardized. The size of the data window used for statistical feature calculation is dynamically adjusted according to the terrain risk coefficient. When the terrain risk coefficient is higher than the preset value, a smaller data window is used to improve real-time performance; otherwise, a larger data window is used to reduce the impact of noise. The standardized traffic index is obtained by weighted calculation of the standardized sound wave velocity value and the standardized tactile friction coefficient value, where the weight coefficient is dynamically allocated by the bidirectional optimization module.

[0056] In this embodiment, the compatibility between the reacquired data and the original data is ensured by a spatiotemporal alignment engine, which injects a global clock signal into the reacquired data to achieve time synchronization; the reacquired data is spatially aligned with the original point cloud coordinate system through visual semantic tags; data conflict resolution rules are established, and when the difference in terrain risk coefficient is greater than a preset value, the twin map is updated based on the reacquired data; when sensor data is not detected in the reacquired data, historical data is used and the confidence level is marked.

[0057] It should be noted that, in addressing the issue of discontinuous positioning within multi-story buildings, a fusion positioning mechanism combining acoustic SLAM and tactile anchor points is employed. When crossing floors: the acoustic array emits 40kHz pulses at 20ms intervals, establishing a vertical point cloud model through reflections from the stairwell walls; the tactile sensor records the foot pressure sequence while climbing the stairs, generating cross-floor motion feature fingerprints; when a floor change is detected, such as a sudden change in the pressure sequence: a dedicated acoustic grid for the stairwell is activated, and a visual sensor assists in recognizing the floor's QR code.

[0058] To overcome the problem of misidentification of complex staircases, a triple verification mechanism of sound wave, touch, and infrared was created: 1. Sound wave ranging: 40kHz pulses penetrate dust / water mist to detect the geometric parameters of the steps; 2. Touch verification: Real-time monitoring by sensor array: edge impact force identifies the leading edge of the steps, and surface texture vibration spectrum distinguishes between metal / stone steps; 3. Infrared compensation: When the sound wave-touch deviation is >5cm: the thermal imaging module (resolution 320×240) is activated to confirm the structure through temperature difference characteristics.

[0059] When inspecting complex terrain, multimodal physical characteristics are fused for perception. For example, tactile friction coefficient combined with infrared temperature can be used to identify oily ground, and acoustic reflectivity combined with tactile vibration spectrum can be used to identify ice surfaces. Visual 3D reconstruction combined with tactile pressure distribution can be used to identify gravel slopes, and different operations can be performed based on different terrains.

[0060] When optimizing the efficiency of dynamic obstacle avoidance, a 16-channel sound source localization array + CPI decision model is deployed to achieve real-time tracking technology, and a hierarchical obstacle avoidance method is adopted to collaboratively generate the optimal avoidance path.

[0061] This invention comprehensively solves the inspection challenges in complex environments by constructing a triple verification mechanism of sound waves, touch, and infrared, combined with a multimodal physical characteristic fusion perception system. For complex staircase recognition, it employs sound pulse detection and tactile edge recognition to collaboratively verify staircase features, combined with thermal imaging temperature difference to lock the staircase mode, completely eliminating gait instability caused by visual misjudgment. For cross-floor positioning continuity interruptions, it utilizes sound wave SLAM and tactile anchor point fusion positioning technology to activate high-density point cloud scanning and visual QR code assistance during floor switching, achieving seamless vertical spatial positioning. For complex terrains such as oil stains and ice surfaces, it dynamically triggers foot adhesion enhancement and bionic spike deployment based on real-time monitoring of tactile friction coefficient and sound wave reflectivity analysis, ensuring passability on special surfaces. For dynamic obstacle avoidance, it relies on a 16-channel sound source positioning array and a CPI hierarchical decision model, using TDOA millisecond-level positioning and multi-robot collaborative obstacle avoidance to achieve trajectory prediction and conflict resolution, ultimately forming a closed-loop system of perception-positioning-decision-execution, effectively solving problems such as staircase misidentification, positioning interruption, terrain instability, and obstacle avoidance failure.

[0062] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for area inspection based on a quadrupedal bionic robot, characterized in that, Includes the following steps: S100: Environmental data is collected synchronously through a multimodal sensing unit. The depth information obtained by the acoustic sensor array and the surface hardness data obtained by the tactile sensor array are fused through a spatiotemporal alignment mechanism to generate a terrain feature vector. S200. Calculate the terrain risk coefficient based on the terrain feature vector, and trigger the stepped anti-slip control chain according to the value of the terrain risk coefficient. S300: The passage index is calculated based on the distance to visual obstacles and the speed of sound wave movement, and the calculation result of the passage index is directly related to the obstacle avoidance decision chain. S400 monitors the attitude stability of S200 and the path deviation of S300 through the closed-loop verification unit. When the attitude stability or path deviation exceeds the tolerance, bidirectional optimization is triggered, and the data re-acquisition process of S100 is restarted in reverse. The calculation weights of the risk coefficient of S200 and the passage index of S300 are adjusted synchronously. The S500 incorporates the optimization results of the S400 into the twin map model, generating an inspection report that includes equipment displacement alarms and joint life predictions, thus completing a closed loop from perception to feedback.

2. The area inspection method for a quadrupedal bionic robot according to claim 1, characterized in that, In S100, the spatiotemporal alignment mechanism includes binding acoustic depth data, tactile hardness data and visual images to the same coordinate system through a spatiotemporal alignment module. Specifically, the acoustic array emits pulse signals at fixed time intervals to generate a depth point cloud; the tactile sensor collects the foot pressure distribution during the pulse interval; when the spatial deviation between the point cloud and the pressure distribution exceeds a set value, the visual sensor is triggered to supplement semantic information for calibration.

3. The area inspection method for a quadrupedal bionic robot according to claim 1, characterized in that, In S200, the terrain risk coefficient is calculated by weighting the ratio of acoustic detection depth to tactile surface hardness. The activation conditions and execution effects of the anti-slip control chain form a closed loop: the foot adhesion force increases stepwise with the increase of the terrain risk coefficient; the stride compression rate and the tactile friction coefficient are proportionally related. When the terrain risk coefficient exceeds the first threshold, the joint verification mechanism of sound wave and tactile sensation is activated to recalibrate the depth data. If the terrain risk factor still exceeds the first threshold after calibration, execute the stride compression command and activate the foot adsorption device.

4. The area inspection method for a quadrupedal bionic robot according to claim 1, characterized in that, In S300, the passage index is jointly determined by the sound wave velocity, the tactile friction coefficient, and the visual obstacle distance. When the passage index falls below a second threshold, a lateral avoidance gait is triggered and a sound wave avoidance signal is emitted. The obstacle avoidance decision chain includes a two-layer feedback mechanism: First layer: Calculate the lateral movement distance based on the traffic index; The second layer: dynamically adjusts the weighting of the passage index calculation based on the actual obstacle avoidance success rate.

5. The area inspection method for a quadrupedal bionic robot according to claim 1, characterized in that, S400 includes the following steps: S410: Real-time monitoring of the quadruped robot's posture stability indicators through a closed-loop verification engine. S420. When the posture stability index is lower than the third threshold, increase the calculation weight of the tactile sensor data in the terrain feature vector, and simultaneously monitor the path tracking deviation data of the quadruped robot. S430. When the path tracking deviation exceeds the fourth threshold, reduce the calculation weight of visual sensor distance data in the passage index. S440. Write the optimized weight parameters into the terrain risk coefficient calculation rules of S200 and the traffic index calculation rules of S300 in real time. S450: Generate new motion control commands based on the updated calculation rules.

6. The area inspection method for a quadrupedal bionic robot according to claim 1, characterized in that, In step S500, the remaining service life of the robot is predicted by real-time acquisition of the joint vibration spectrum. When the remaining service life is less than the preset time, the location of high-risk equipment is marked in the digital twin map, and the inspection path is optimized to prioritize the coverage of maintenance stations. At the same time, a maintenance alarm is generated and associated with the equipment displacement data.

7. The area inspection method for a quadrupedal bionic robot according to claim 3, characterized in that, The normalization of the sound wave velocity and tactile friction coefficient adopts a dynamic standardization method. Based on the statistical characteristics of historical data, the original sound wave velocity and tactile friction coefficient are standardized. The size of the data window used for statistical feature calculation is dynamically adjusted according to the terrain risk coefficient. When the terrain risk coefficient is higher than the preset value, a smaller data window is used to improve real-time performance; otherwise, a larger data window is used to reduce the impact of noise. The standardized traffic index is obtained by weighted calculation of the standardized sound wave velocity value and the standardized tactile friction coefficient value, where the weight coefficient is dynamically allocated by the bidirectional optimization module.

8. The area inspection method for a quadrupedal bionic robot according to claim 1, characterized in that, The compatibility between the reacquisition data and the original data is ensured by a spatiotemporal alignment engine, which injects a global clock signal into the reacquisition data to achieve time synchronization; the reacquisition data is spatially aligned with the original point cloud coordinate system through visual semantic labels; data conflict resolution rules are established, and when the difference in terrain risk coefficient is greater than a preset value, the twin map is updated based on the reacquisition data; when sensor data is not detected in the reacquisition, historical data is used and the confidence level is marked.

9. A region inspection system for a quadrupedal bionic robot, the system being used to implement the region inspection method of the quadrupedal bionic robot according to any one of claims 1-8, enabling the quadrupedal bionic robot to complete region inspection operations, characterized in that, include: Multimodal perception unit, intelligent decision-making unit, closed-loop verification unit, and twin map management unit.

10. The quadrupedal bionic robot area inspection system according to claim 9, characterized in that, The multimodal sensing unit includes an integrated acoustic and tactile module, which uses a piezoelectric ceramic emitter and a PVDF tactile sensor on the same substrate to output a data stream bound to the same coordinate system; the visual compensation module is activated when the deviation between the acoustic and tactile data is greater than a threshold. The intelligent decision-making unit includes a risk control chain module, which receives terrain feature vectors and outputs gait commands linked to terrain risk coefficients; and an obstacle avoidance decision chain module, which receives passage indexes and outputs control signals linked to avoidance distances.

Images