Underwater soft bottom walking robot anti-trapping system based on acoustic perception and adaptive model predictive control

Abstract

The application discloses an underwater soft bottom walking robot anti-trapping system based on acoustic perception and adaptive model predictive control, and belongs to the technical field of underwater robot control, which comprises a multi-frequency acoustic impedance perception module, a collapse probability field construction module, a state estimation module, an adaptive model predictive controller (AMPC), a multi-actuator collaborative impedance controller, a left track driving module, a right track driving module, a propeller driving module and a gravity center adjusting mechanism driving module. The multi-frequency acoustic impedance perception module is connected with the collapse probability field construction module and the state estimation module respectively, the collapse probability field construction module and the state estimation module are connected with two input ends of the AMPC, the AMPC is connected with each actuator through the multi-actuator collaborative impedance controller, and the actuator feedback and sensor data input state estimation module. The application adopts the above system, realizes multi-actuator collaborative anti-trapping, and significantly improves the passability, safety and energy efficiency of the underwater robot in the soft bottom environment.

Description

Technical Field

[0001] This invention relates to the field of underwater robot control technology, and in particular to an anti-sinking system for underwater soft-bottom walking robots based on acoustic perception and adaptive model predictive control. Background Technology

[0002] When underwater robots navigate in soft seabed environments such as silt and sediment, they often sink due to insufficient seabed bearing capacity, leading to mission interruptions, increased energy consumption, and difficulties in escaping. Currently, existing technologies mainly employ the following solutions: (1) Control scheme based on sonar terrain recognition: Typical examples include BlueROV2 and RemusAUV, which obtain terrain elevation through multibeam sonar and adjust the thruster output by combining PID controller. This scheme can only identify the terrain outline and cannot sense the bearing capacity of the substrate. Therefore, it may still sink in on a visually flat but actually soft substrate.

[0003] (2) Tracked crawling robot solution: Typical examples include the Alvin manned submersible and the Sea Crawler series. It adopts a wide track design to reduce the ground pressure and avoids sinking by adjusting the track speed. The control parameters of this solution are fixed and cannot be dynamically adjusted according to changes in the seabed, resulting in poor adaptability.

[0004] (3) Obstacle avoidance schemes based on optical vision: Typical examples include WHOI's Nereus and SeaXplorer systems, which use high-definition cameras to collect terrain images and combine them with machine learning to identify dangerous areas. This scheme is prone to failure in turbid underwater environments and can only identify surface features, unable to determine the underground bearing capacity of the substrate.

[0005] In summary, existing technologies suffer from the following shortcomings: insufficient sensing capabilities, failing to quantify key mechanical parameters such as seabed bearing capacity and viscoelastic modulus in real time; rigid control strategies, lacking a response mechanism to dynamic changes in the seabed; lagging path planning, typically initiating escape operations only after sinking, lacking prior warning; poor actuator coordination, with a lack of unified control among propellers, tracks, and attitude adjustment mechanisms; and a lack of energy efficiency optimization, resulting in significantly increased energy consumption in soft seabed environments. Therefore, there is an urgent need for an anti-sinking system capable of real-time sensing of seabed mechanical properties, dynamic planning of avoidance paths, and adaptive adjustment of multiple actuators. Summary of the Invention

[0006] The purpose of this invention is to provide an anti-sinking system for underwater soft-bottom walking robots based on acoustic perception and adaptive model predictive control. This system can sense the bearing capacity of the seabed in real time, construct a sinking probability field, dynamically plan avoidance paths, adaptively adjust the impedance parameters of multiple actuators, provide early warning in high-risk areas, avoid sinking, and improve passability and energy efficiency.

[0007] To achieve the above objectives, this invention provides an anti-sinking system for underwater soft-bottom walking robots based on acoustic sensing and adaptive model predictive control, comprising: Multi-frequency acoustic impedance sensing module, trap probability field construction module, state estimation module, adaptive model predictive controller (AMPC), multi-actuator cooperative impedance controller, left track drive module, right track drive module, thruster drive module, center of gravity adjustment mechanism drive module; The output of the multi-frequency acoustic impedance sensing module is connected to the input of the trapped probability field construction module and the first input of the state estimation module, respectively. The output of the trap probability field building module is connected to the first input of the adaptive model prediction controller (AMPC); The output of the state estimation module is connected to the second input of the adaptive model prediction controller (AMPC); The output of the adaptive model predictive controller (AMPC) is connected to the input of the multi-actuator coordinated impedance controller; The output of the multi-actuator coordinated impedance controller is connected to the left track drive module, the right track drive module, the thruster drive module, and the center of gravity adjustment mechanism drive module, respectively. The execution status feedback of the left track drive module, right track drive module, thruster drive module, and center of gravity adjustment mechanism drive module is connected to the second input terminal of the state estimation module.

[0008] Preferably, the multi-frequency acoustic impedance sensing module includes a 28kHz and 120kHz dual-frequency acoustic wave transmitting array and a 16-element sound pressure receiving array.

[0009] Preferably, it also includes a bottom contact pressure sensor, the output of which is connected to the multi-frequency acoustic impedance sensing module. The bottom contact pressure sensor is used to calibrate the substrate bearing capacity obtained by the multi-frequency acoustic impedance sensing module through acoustic inversion.

[0010] Preferably, the state estimation module is used to generate the state vector. And output to the adaptive model prediction controller (AMPC); Wherein, the state vector ; The adaptive model predictive controller (AMPC) is configured to receive the state vector. And output control input vector ; Among them, the control input vector .

[0011] Preferably, the Adaptive Model Predictive Controller (AMPC) incorporates a rolling optimization objective function: ; in, This represents the total cost function value. Indicates the length of the prediction time domain. Indicates the index of prediction steps. This represents the risk weighting coefficient. This represents the energy consumption weighting coefficient. This represents the path deviation weighting coefficient; Indicates the first The risk of falling into a trap, and ,in, Indicates the bearing capacity of the substrate. Indicates the first The depth of the subsidence, Indicates the safe subsidence depth; Indicates the first The energy consumption item of the step, and ,in, This indicates the change in the rotational speed of the left track. This indicates the change in the rotational speed of the right track. This indicates the change in thrust of the propeller. This indicates the change in the center of gravity adjustment amount; Indicates the first The path deviation term of the step, and ,in, Indicates the first The actual position of the robot. Indicates the first The reference path location for the step. Represents the square of the Euclidean norm; The system also includes a weighted adaptive adjustment module, which is used to dynamically adjust the weighted adaptive adjustment based on the current bearing capacity of the substrate. , , The value of .

[0012] Preferably, the collapse probability field construction module maps the subsurface bearing capacity data output by the multi-frequency acoustic impedance sensing module into a two-dimensional gridded collapse probability map, wherein the collapse probability value of each grid cell is inversely proportional to the subsurface bearing capacity at the corresponding location, and when the subsurface bearing capacity is lower than 15 kPa, the corresponding grid cell is marked as a high-risk area.

[0013] Preferably, the multi-actuator coordinated impedance controller has a built-in virtual impedance model: ; in, Represents virtual resistance. This represents the stiffness coefficient, reflecting the system's response strength to deviations in settlement depth. Indicates the depth of the target subsidence. Indicates the actual subsidence depth. This represents the damping coefficient, which reflects the intensity of the system's response to the settlement velocity; Indicates the rate of subsidence. and Configured to increase when the substrate bearing capacity decreases. and The value increases as the track slip ratio increases. The value; The multi-actuator coordinated impedance controller also includes a priority arbitration unit, which determines the execution priority among the left track drive module, right track drive module, thruster drive module and center of gravity adjustment mechanism drive module based on the current state.

[0014] Preferably, it also includes a mode switching module, which switches the system from normal travel mode to emergency escape mode when the current grid cell's collapse probability output by the collapse probability field construction module exceeds 85%. In emergency escape mode, the path deviation weights in the adaptive model predictive controller (AMPC) The risk weight was automatically reduced to 0.2–0.3. It was automatically raised to above 0.9.

[0015] Preferably, the left track drive module and the right track drive module each include an independent DC motor and a photoelectric encoder slip ratio sensor; The thruster drive module includes a front thruster and a rear thruster; The center of gravity adjustment mechanism drive module includes a servo motor, a slide rail, and a movable counterweight.

[0016] Preferably, it also includes an IMU attitude sensor, a slip rate sensor, and a depth sensor; The IMU attitude sensor is used to measure the robot's pitch and roll angles, the slip ratio sensor is used to measure the track slip ratio, and the depth sensor is used to measure the actual sinking depth. The outputs of the IMU attitude sensor, slip rate sensor, and depth sensor are all connected to the third input of the state estimation module.

[0017] Therefore, the underwater soft-bottom walking robot anti-sinking system based on acoustic perception and adaptive model predictive control, as described above, has the following beneficial effects: (1) Real-time sensing of the mechanical properties of the substrate: Through the dual-frequency acoustic impedance sensing module, the substrate bearing capacity, stiffness and viscoelastic modulus can be quantitatively inverted, which makes up for the shortcomings of existing technologies that can only identify terrain or surface features, and provides accurate mechanical basis for anti-sinking control.

[0018] (2) Adaptive anti-sinking and risk avoidance planning: Construct a sinking probability field and combine it with an adaptive model predictive controller. Adjust the risk, energy consumption and path deviation weights dynamically according to the bottom bearing capacity. Provide early warning and plan risk avoidance paths in high-risk areas, and transform from passive escape to active anti-sinking.

[0019] (3) Multi-actuator coordination and energy efficiency optimization: The virtual impedance model and priority arbitration mechanism are adopted to coordinate the tracks, propellers and center of gravity adjustment mechanism, and the stiffness / damping parameters are adaptively adjusted according to the bottom and slip ratio; at the same time, the energy consumption weight is dynamically optimized according to the battery power, which significantly improves the soft bottom passability and reduces energy consumption.

[0020] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0021] Figure 1 This is a diagram of the underwater soft-bottom walking robot anti-sinking system architecture based on acoustic perception and adaptive model predictive control according to the present invention. Figure 2 This is a flowchart of the acoustic impedance sensing processing of the multi-frequency acoustic impedance sensing module of the present invention. Figure 3 This is an internal computational block diagram of the adaptive model predictive controller (AMPC) of this invention; Figure 4 This is a schematic diagram of the trap probability field construction generated by the trap probability field construction module of the present invention; Figure 5 This is the control logic diagram of the multi-actuator cooperative impedance controller of the present invention; Figure 6 This is a schematic diagram (top view) of the hardware layout of the underwater robot of the present invention. Figure 7 This is the system control flowchart of the present invention (including the switching between normal mode and emergency escape mode). Detailed Implementation

[0022] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0024] Example 1 This embodiment provides an anti-sinking system for an underwater soft-bottom walking robot based on acoustic perception and adaptive model predictive control. Its hardware composition is shown in Table 1.

[0025] Table 1 System Hardware Composition

[0026] The installation positions of the above modules on the robot body are as follows: Figure 6 (Hardware layout diagram of the underwater robot) is shown. The forward acoustic array of the multi-frequency acoustic impedance sensing module is mounted on the front waterproof shell of the robot, while the lateral acoustic arrays are symmetrically mounted on the left and right sides of the robot body. The bottom pressure sensor is mounted below the tracks, in contact with the seabed. The left and right track drive modules are independent. The front and rear thrusters are symmetrically mounted and can be vertically adjusted. The counterweight of the center of gravity adjustment mechanism drive module slides on a guide rail via a servo motor. The control computing unit is installed inside the waterproof chamber and communicates with each sensor and actuator via a watertight connector.

[0027] The system architecture of this embodiment is as follows: Figure 1 As shown. After the system starts, perform the following operations in sequence: Acoustic Sensing: The multi-frequency acoustic impedance sensing module emits dual-frequency sound waves of 28kHz and 120kHz in real time. The reflected signals are received through a 16-element sound pressure receiving array. Sound pressure and particle velocity are extracted, and the acoustic impedance of the substrate is calculated. Then, dual-frequency fusion processing is performed to invert the substrate stiffness, viscoelastic modulus, and bearing capacity. The internal processing flow of this module is as follows: Figure 2 As shown.

[0028] State estimation: The state estimation module receives three inputs: First input terminal: Substrate bearing capacity output by the multi-frequency acoustic impedance sensing module; Second input: feedback on the actual execution status of the left track drive module, right track drive module, thruster drive module, and center of gravity adjustment mechanism drive module; The third input terminal includes: IMU attitude sensor (pitch angle, roll angle), slip ratio sensor (track slip ratio), and depth sensor (actual sinking depth).

[0029] The state estimation module integrates the above data to generate a state vector. .

[0030] Risk Assessment: The collapse probability field construction module receives the subsurface bearing capacity data output by the multi-frequency acoustic impedance sensing module and maps it into a two-dimensional gridded collapse probability map (e.g., Figure 4 As shown in the figure, the collapse probability of each grid cell is inversely proportional to the bearing capacity. When the bearing capacity is below 15 kPa, the corresponding grid cell is marked as a high-risk area (red). This probability map is input to the adaptive model predictive controller (AMPC) in real time.

[0031] Adaptive Predictive Control: AMPC simultaneously receives the trap probability field (first input) and the state vector (second input), and uses its built-in rolling optimization objective function to predict the optimal control sequence over the next N steps. The internal computation flow of AMPC is as follows... Figure 3 As shown. The optimal control input vector at the current time is obtained by solving. It then outputs its data to a multi-actuator coordinated impedance controller.

[0032] Cooperative impedance control: The multi-actuator cooperative impedance controller receives the control vector from the AMPC and performs the following operations (such as...). Figure 5 (as shown) Calculate virtual impedance: based on the virtual impedance model Calculate the virtual impedance. Wherein, For the target subsidence depth, This represents the actual subsidence depth. This refers to the subsidence rate. (Stiffness coefficient) and The damping coefficient is dynamically adjusted based on the current bearing capacity of the subgrade and the slip ratio of the tracks.

[0033] In this embodiment, the stiffness coefficient The range of values ​​is 0 < ≤1000 N / m, damping coefficient The range of values ​​is 0 < ≤200 Ns / m. It adaptively adjusts based on the bottom bearing capacity and track slip ratio. It adaptively adjusts based on track slip ratio and sinking acceleration.

[0034] Priority arbitration: such as Figure 5 As shown, the multi-actuator cooperative impedance controller also includes a priority arbitration unit, which determines the execution priority among the tracks, propellers, and center of gravity adjustment mechanisms based on the current state.

[0035] Force distribution and execution: According to the arbitration result, the virtual impedance force and AMPC control vector are distributed to the left track drive module, right track drive module, thruster drive module and center of gravity adjustment mechanism drive module, and the corresponding actions are executed.

[0036] After execution, the actual state of each actuator is fed back to the second input of the state estimation module.

[0037] Execution and Feedback: The left track drive module, right track drive module, thruster drive module, and center of gravity adjustment mechanism drive module operate according to the assigned instructions. The actual state of each actuator (speed, thrust, center of gravity position) is fed back to the second input of the state estimation module, forming a closed-loop control.

[0038] Mode Switching: The system also includes a mode switching module. This module monitors the fall probability of the current grid cell output by the fall probability field construction module in real time. When the fall probability exceeds 85%, the mode switching module switches the system from normal travel mode to emergency escape mode; after successful escape, it switches back to normal travel mode. The control flowchart of the entire system is as follows: Figure 7 As shown.

[0039] In this embodiment, AMPC employs a rolling optimization approach, and its cost function is: ; in, This represents the total cost function value. This indicates the prediction time domain length (5 in this embodiment). Indicates the index of prediction steps. This represents the risk weighting coefficient. This represents the energy consumption weighting coefficient. The above three weight coefficients are dynamically adjusted by the weight adaptive adjustment module. The values ​​of the three weight coefficients are all greater than 0 and less than 1. Indicates the first The risk of falling into a trap, and ,in, Indicates the bearing capacity of the substrate. Indicates the first The depth of the subsidence, Indicates the safe settlement depth (unit: cm, set by the system or calculated by the bottom sediment model).

[0040] Indicates the first The energy consumption item of the step, and ,in, Indicates the change in left track speed. This indicates the change in the right track rotation speed (in rpm, obtained from feedback by the photoelectric encoder). This indicates the change in thrust of the thruster (in N, measured by the thruster force sensor). This indicates the change in center of gravity adjustment (unit: kg·m, obtained from feedback from the servo motor). Indicates the first The path deviation term of the step, and ,in, Indicates the first The actual position (coordinates) of the robot (Unit: m, obtained by INS and DVL fusion positioning) Indicates the first Reference path location (coordinates) for each step (Unit: m, generated by the path planning module using the A or DLite algorithm). Represents the square of the Euclidean norm; The weight adaptive adjustment module dynamically adjusts the load based on the current substrate bearing capacity and the remaining battery capacity. , , The specific adjustment logic is as follows: (Risk weight): In high-risk areas (e.g., bearing capacity <15 kPa), Automatically increases to near 1, prioritizing reducing the probability of collapse; in low-risk areas (e.g., bearing capacity > 30 kPa), It can be reduced to 0.3~0.5, allowing for some risk in order to improve efficiency.

[0041] (Energy consumption weight): When the battery charge is >80%, =0.5, allowing for higher energy consumption; when battery charge is <30%, Automatically increases to 0.8~1.0, prioritizing energy saving.

[0042] (Path Deviation Weight): In task path planning, =0.7~1.0, emphasizing path tracking accuracy; in emergency escape mode, Automatically reduce to 0.2~0.3, allowing temporary path deviation for emergency escape.

[0043] Table 2 provides example weight configurations for different task scenarios.

[0044] Table 2 Examples of weight coefficients in different task scenarios

[0045] To better understand the collaborative process of each module, the following example illustrates how an underwater robot equipped with this system performs scientific research in a seabed silt area.

[0046] The robot moves forward at a speed of 0.5 m / s. The multi-frequency acoustic impedance sensing module emits dual-frequency sound waves of 28kHz and 120kHz in real time, receives the echoes, and calculates that the bearing capacity of the substrate within 2 m ahead is 12 kPa (below the 15 kPa threshold), classifying it as a high-risk area. The state estimation module simultaneously collects IMU data (pitch angle 2°, roll angle 1°), slip ratio sensor (left track slip ratio 8%), depth sensor (current subsidence depth 3 cm), and actuator feedback to form a state vector. The collapse probability field construction module receives bearing capacity data, generates a two-dimensional grid map, marks the grid with a bearing capacity of 12 kPa in red (collapse probability 0.7), and inputs it into AMPC. AMPC receives the state vector and risk map, predicting that if it continues straight ahead within the next 3 seconds, it will enter a high-risk zone. At this time, because the bearing capacity is below 15 kPa, the weight adaptive adjustment module will... Rising to 0.9 =0.6, =0.4. The optimized solution outputs the following control commands: left track speed reduced by 10%, right track speed reduced by 5%, rear thruster thrust increased by 20 N, and center of gravity shifted rearward by 2 cm. The multi-actuator coordinated impedance controller receives these commands and first calculates the virtual impedance force based on the virtual impedance model. The priority arbitration unit determines the execution priority based on the current state (e.g., ...). Figure 5 As shown in the diagram, the instructions are distributed to each actuator. After execution, the robot's actual sinking depth decreases to 2.5 cm, and the slip rate decreases to 5%, which is fed back to the state estimation module. In the next cycle, if the load-bearing capacity further decreases to 8 kPa, and the sinking probability field calculation shows that the sinking probability of the current grid cell exceeds 85%, the mode switching module automatically switches the system to emergency escape mode. In emergency escape mode, AMPC will... Rising to 0.95 When the impedance is reduced to 0.3 (see Table 2), a temporary deviation from the original path is allowed. The multi-actuator collaborative impedance controller adjusts the priority strategy, and the robot switches back to normal travel mode after successfully escaping the obstacle.

[0047] Therefore, this invention employs the aforementioned underwater soft-bottom walking robot anti-sinking system based on acoustic perception and adaptive model predictive control. By using a multi-frequency acoustic impedance sensing module to quantify the seabed stiffness, viscoelastic modulus, and bearing capacity in real time, and combining this with a sinking probability field and adaptive model predictive control, it achieves active anti-sinking protection in soft-bottom environments. Compared to existing technologies, this invention significantly improves the robot's mobility and safety in soft terrains such as silt and sediment, reducing the risk of sinking. Simultaneously, through adaptive weight adjustment and multi-actuator collaborative impedance control, it optimizes energy consumption and extends underwater operation time. In high-risk areas, this system can provide early warnings and automatically switch to an emergency escape mode, effectively avoiding mission interruptions and escape difficulties. It is suitable for complex tasks such as deep-sea exploration and subsea pipeline inspection.

[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. An anti-sinking system for underwater soft-bottom walking robots based on acoustic perception and adaptive model predictive control, characterized in that, include: Multi-frequency acoustic impedance sensing module, trap probability field construction module, state estimation module, adaptive model prediction controller, multi-actuator cooperative impedance controller, left track drive module, right track drive module, thruster drive module, center of gravity adjustment mechanism drive module; The output of the multi-frequency acoustic impedance sensing module is connected to the input of the trapped probability field construction module and the first input of the state estimation module, respectively. The output of the trap probability field construction module is connected to the first input of the adaptive model prediction controller. The output of the state estimation module is connected to the second input of the adaptive model prediction controller; The output of the adaptive model predictive controller is connected to the input of the multi-actuator coordinated impedance controller; The output of the multi-actuator coordinated impedance controller is connected to the left track drive module, the right track drive module, the thruster drive module, and the center of gravity adjustment mechanism drive module, respectively. The execution status feedback of the left track drive module, right track drive module, thruster drive module, and center of gravity adjustment mechanism drive module is connected to the second input terminal of the state estimation module.

2. The underwater soft-bottom walking robot anti-sinking system based on acoustic perception and adaptive model predictive control according to claim 1, characterized in that, The multi-frequency acoustic impedance sensing module includes a 28kHz and 120kHz dual-frequency acoustic wave transmitting array and a 16-element sound pressure receiving array.

3. The underwater soft-bottom walking robot anti-sinking system based on acoustic perception and adaptive model predictive control according to claim 1, characterized in that, It also includes a bottom contact pressure sensor, the output of which is connected to a multi-frequency acoustic impedance sensing module. The bottom contact pressure sensor is used to calibrate the substrate bearing capacity obtained by the multi-frequency acoustic impedance sensing module through acoustic inversion.

4. The underwater soft-bottom walking robot anti-sinking system based on acoustic perception and adaptive model predictive control according to claim 1, characterized in that, The state estimation module is used to generate state vectors. And output to the adaptive model prediction controller; Wherein, the state vector ; The adaptive model predictive controller is configured to receive the state vector. And output control input vector ; Among them, the control input vector .

5. The underwater soft-bottom walking robot anti-sinking system based on acoustic perception and adaptive model predictive control according to claim 4, characterized in that, The adaptive model predictive controller has a built-in rolling optimization objective function: ； in, This represents the total cost function value. Indicates the length of the prediction time domain. Indicates the index of prediction steps. This represents the risk weighting coefficient. This represents the energy consumption weighting coefficient. This represents the path deviation weighting coefficient; Indicates the first The risk of falling into a trap, and ,in, Indicates the bearing capacity of the substrate. Indicates the first The depth of the subsidence, Indicates the safe subsidence depth; Indicates the first The energy consumption item of the step, and ,in, This indicates the change in the rotational speed of the left track. This indicates the change in the rotational speed of the right track. This indicates the change in thrust of the propeller. This indicates the change in the center of gravity adjustment amount; Indicates the first The path deviation term of the step, and ,in, Indicates the first The actual position of the robot. Indicates the first The reference path location for the step. Represents the square of the Euclidean norm; The system also includes a weighted adaptive adjustment module, which is used to dynamically adjust the weighted adaptive adjustment based on the current bearing capacity of the substrate. , , The value of .

6. The underwater soft-bottom walking robot anti-sinking system based on acoustic perception and adaptive model predictive control according to claim 5, characterized in that, The collapse probability field construction module maps the bottom bearing capacity data output by the multi-frequency acoustic impedance sensing module into a two-dimensional gridded collapse probability map. The collapse probability value of each grid cell is inversely proportional to the bottom bearing capacity at the corresponding location. When the bottom bearing capacity is lower than 15 kPa, the corresponding grid cell is marked as a high-risk area.

7. The underwater soft-bottom walking robot anti-sinking system based on acoustic perception and adaptive model predictive control according to claim 1, characterized in that, The multi-actuator coordinated impedance controller has a built-in virtual impedance model: ； in, Represents virtual resistance. This represents the stiffness coefficient, reflecting the system's response strength to deviations in settlement depth. Indicates the depth of the target subsidence. Indicates the actual subsidence depth. This represents the damping coefficient, which reflects the intensity of the system's response to the settlement velocity; Indicates the rate of subsidence. and Configured to increase when the substrate bearing capacity decreases. and The value increases as the track slip ratio increases. The value; The multi-actuator coordinated impedance controller also includes a priority arbitration unit, which determines the execution priority among the left track drive module, right track drive module, thruster drive module and center of gravity adjustment mechanism drive module based on the current state.

8. The underwater soft-bottom walking robot anti-sinking system based on acoustic perception and adaptive model predictive control according to claim 6, characterized in that, It also includes a mode switching module, which switches the system from normal travel mode to emergency escape mode when the current grid cell's collapse probability exceeds 85%, as output by the collapse probability field construction module. In emergency escape mode, the path deviation weights in the adaptive model predictive controller The risk weight was automatically reduced to 0.2–0.

3. It was automatically raised to above 0.

9.

9. The underwater soft-bottom walking robot anti-sinking system based on acoustic perception and adaptive model predictive control according to claim 1, characterized in that, Each of the left track drive module and the right track drive module includes an independent DC motor and a photoelectric encoder slip ratio sensor. The thruster drive module includes a front thruster and a rear thruster; The center of gravity adjustment mechanism drive module includes a servo motor, a slide rail, and a movable counterweight.

10. The underwater soft-bottom walking robot anti-sinking system based on acoustic perception and adaptive model predictive control according to claim 1, characterized in that, It also includes an IMU attitude sensor, a slip rate sensor, and a depth sensor; The IMU attitude sensor is used to measure the robot's pitch and roll angles, the slip ratio sensor is used to measure the track slip ratio, and the depth sensor is used to measure the actual sinking depth. The outputs of the IMU attitude sensor, slip rate sensor, and depth sensor are all connected to the third input of the state estimation module.

Images