An amphibious robot motion optimization method, system and device
By constructing a control parameter reference library and a fuzzy rule library based on historical data, the problem of poor motion control performance of amphibious robots in complex environments was solved, and higher posture stability and motion accuracy were achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING INST OF TECH
- Filing Date
- 2026-05-22
- Publication Date
- 2026-06-19
AI Technical Summary
The motion control performance of existing amphibious robots is poor, mainly because the rule base of fuzzy PID control technology cannot effectively adapt to complex environments, resulting in rule base dimensionality disaster and rule conflicts, which affect posture stability and motion accuracy.
By acquiring historical motion sample information of amphibious robots in different motion scenarios, frequently occurring and effective combinations of control parameters are selected, a distribution reference range is constructed, and adaptive control parameter matching is performed using FP-trees and fuzzy rule bases to avoid rule conflicts.
It improves the motion control performance of amphibious robots, enhances their posture stability and motion accuracy in complex environments, and achieves adaptive control strategy matching.
Smart Images
Figure CN122239726A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of program-controlled robot motion technology, specifically to a motion optimization method, system, and device for an amphibious robot. Background Technology
[0002] Amphibious robots, as intelligent equipment capable of performing tasks across media, have become core tools in fields such as rescue and disaster relief, military reconnaissance, and marine exploration. As mission scenarios expand from single environments to complex environments, the accuracy, robustness, and environmental adaptability of amphibious robot motion control have become core technical indicators.
[0003] Currently, motion control of amphibious robots is usually based on fuzzy PID control technology. However, the existing fuzzy PID rule base is mostly built based on expert experience in a single scenario of pure land or pure water. It does not take into account the impact of environmental changes during scenario transitions on the motion characteristics of amphibious robots. Furthermore, introducing different rule bases for different environments can lead to the disaster of dimensionality in the rule base, resulting in numerous and complex rules, and even rule conflicts. This restricts the posture stability and motion accuracy of the robot in cross-media movement, resulting in poor motion control performance for amphibious robots. Summary of the Invention
[0004] To address the technical problem of poor motion control performance for amphibious robots, the present invention aims to provide a motion optimization method, system, and apparatus for amphibious robots. The specific technical solution adopted is as follows: A motion optimization method for an amphibious robot, the method comprising: Acquire motion sample information of the amphibious robot in each motion scenario. The motion sample information includes road condition data, control parameter combinations and control effect parameters at each historical motion moment. The motion scenarios include various land scenarios, water scenarios and water-land transition scenarios. Based on the distribution characteristics of control parameter combinations at all historical motion times, and the corresponding control effect parameters for each control parameter combination, all reference control parameter combinations are selected from all control parameter combinations; based on the road condition data and control effect parameters corresponding to each reference control parameter combination at all historical motion times, the distribution reference range of road condition data is obtained. At the current moment of motion, based on the distribution reference range of the current road condition data of the amphibious robot, the control parameters of the amphibious robot are determined and its motion is controlled.
[0005] Furthermore, the road condition data includes at least all motion sensor data and all environmental sensor data during the movement of the amphibious robot.
[0006] Furthermore, the method for obtaining the combination of control parameters includes: The proportional, integral, and derivative parameters of the PID controller of the amphibious robot are used as control parameters. Based on the numerical distribution of the proportional parameters at all historical motion moments, a preset number of initial distribution intervals for the proportional parameters are defined; the preset number is the same as the number of motion scenes; the midpoint of each initial distribution interval is used as the cluster center to cluster all proportional parameters, and the distribution interval of the proportional parameters is determined based on the clustering results. Based on the numerical distribution of integral parameters at all historical motion moments, high-risk and low-risk distribution intervals for integral saturation are defined; based on the numerical distribution of differential parameters at all historical motion moments, high-risk and low-risk distribution intervals for noise amplification are defined. At each historical moment of motion, the combination of control parameters is determined based on the distribution range of each control parameter.
[0007] Furthermore, the method for obtaining the control effect parameters includes: At each historical moment of motion, the dynamic error, steady-state error, and steady-state convergence time of the amphibious robot are acquired. Based on preset weights, the negative correlation normalization results of the dynamic error, the steady-state error, and the steady-state convergence time are weighted respectively, and the weighted sum is used as the control effect parameter.
[0008] Furthermore, the method for obtaining the reference control parameter combination includes: An FP tree is constructed based on the combination of control parameters at historical motion moments, where each node in the FP tree is a single combination parameter of the control parameter combination. Based on the control effect parameters and frequency of occurrence of each control parameter combination, obtain the reference index of each individual combination parameter in the FP tree; Based on the reference index and the preset minimum support, the FP tree is pruned, and all corresponding control parameter combinations in the pruned FP tree are used as reference control parameter combinations.
[0009] Furthermore, the method for obtaining the reference index includes: The ratio of the frequency of occurrence of each combination of control parameters to the maximum frequency of occurrence among all combinations of control parameters is used as the combination frequency parameter. Under each combination of control parameters, for each individual combination parameter in the FP tree, the control effect parameter of each combination of control parameters in which it is located is used to weight and sum the corresponding combination of frequent parameters to obtain the sub-reference coefficient; The reference index is obtained by combining the sub-reference coefficients under all combinations of control parameters.
[0010] Furthermore, the method for obtaining the distribution reference range includes: Using the control effect parameters at each historical motion moment as weights, and based on the differences between road condition data at different historical motion moments, weighted mean drift clustering is performed on the road condition data at all historical motion moments corresponding to each combination of reference control parameters, and the distribution reference range of the road condition data is determined based on the clustering results.
[0011] Furthermore, methods for determining the control parameters of amphibious robots include: The initial control parameters are determined by combining the reference control parameters corresponding to the distribution reference range of the current road condition data; the initial control parameters are then fine-tuned based on a preset fuzzy rule base to obtain the final control parameters.
[0012] A motion optimization system for an amphibious robot, the system comprising: Historical data acquisition module: used to acquire motion sample information of amphibious robot in each motion scenario. The motion sample information includes road condition data, control parameter combinations and control effect parameters at each historical motion moment. The motion scenarios include various land scenarios, water scenarios and water-land transition scenarios. Historical data analysis module: Based on the distribution characteristics of control parameter combinations at all historical motion times, and the corresponding control effect parameters for each control parameter combination, it filters out all reference control parameter combinations from all control parameter combinations; and obtains the distribution reference range of road condition data based on the road condition data and control effect parameters corresponding to each reference control parameter combination at all historical motion times. Motion analysis and control module: used to determine the control parameters and perform motion control on the amphibious robot based on the distribution reference range of the current road condition data at the current moment of motion.
[0013] A motion optimization device for an amphibious robot includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the motion optimization method for the amphibious robot.
[0014] The present invention has the following beneficial effects: This invention first acquires motion sample information of an amphibious robot in each motion scenario. This motion sample information includes road condition data, control parameter combinations, and control effect parameters for each historical motion moment, providing historical reference for subsequent motion optimization analysis. Then, based on the distribution characteristics of control parameter combinations across all historical motion moments and the corresponding control effect parameters for each combination, all frequently occurring and effective historical control strategy reference control parameter combinations are selected from all combinations, providing effective historical strategy references for subsequent motion optimization. Next, based on the road condition data and control effect parameters corresponding to each reference control parameter combination across all historical motion moments, a distribution reference range for road condition data is obtained, preparing for subsequent matching and application of historical control strategies based on road condition data. Finally, at the current motion moment, based on the distribution reference range of the amphibious robot's current road condition data, the control parameters of the amphibious robot are determined, and its motion is controlled. This invention, based on the effective references provided by the amphibious robot's historical motion data and control strategies, can adaptively match control strategies based on the current motion situation, avoiding rule conflicts and improving the motion control effect of the amphibious robot. Attached Figure Description
[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0016] Figure 1 A flowchart illustrating a motion optimization method for an amphibious robot according to an embodiment of the present invention; Figure 2 A flowchart illustrating a method for obtaining a combination of reference control parameters according to an embodiment of the present invention; Figure 3 This is an example diagram of a preset fuzzy rule base provided by the present invention. Detailed Implementation
[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation methods, structures, features, and effects of a motion optimization method, system, and apparatus for an amphibious robot proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.
[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0019] The following description, in conjunction with the accompanying drawings, details a specific scheme for a motion optimization method, system, and device for an amphibious robot provided by the present invention.
[0020] Please see Figure 1 The diagram illustrates a flowchart of a motion optimization method for an amphibious robot according to an embodiment of the present invention, specifically including: Step S1: Obtain motion sample information of the amphibious robot in each motion scenario. The motion sample information includes road condition data, control parameter combinations and control effect parameters for each historical motion moment. The motion scenarios include various land scenarios, water scenarios and water-land transition scenarios.
[0021] One embodiment of the present invention first collects a large amount of historical motion process data of amphibious robots in different motion scenarios, and uses the historical motion data as motion sample information to provide historical reference for subsequent analysis and motion optimization; the motion sample information needs to include the amphibious robot's road condition data, control parameter combinations and control effect parameters; the motion scenarios include various land scenarios, water scenarios and water-land transition scenarios.
[0022] The land scenes mainly include flat hard surfaces such as asphalt, concrete, and smooth dirt roads; undulating and sloping terrain such as gentle slopes, steep slopes, and hilly areas; soft and unstructured terrain such as sand, mud, grassland, and snow; and obstacle terrain such as gravel roads, areas with steps or obstacles. The water-land transition operation scenarios mainly include the water entry process (entering the water from the shore, where buoyancy gradually replaces the supporting force), the water exit process (climbing from the water to the shore, where buoyancy rapidly decreases), and wading in shallow water (the robot is not fully afloat, and the bottom resistance is mixed with the water surface wave resistance). The underwater operation scenarios mainly include calm waters such as indoor pools and lakes in windless weather; slow-flowing areas such as rivers and canals with stable water flow; windy and wave areas such as open waters with surface waves caused by wind; shallow and turbulent areas such as nearshore areas, estuaries, and areas with underwater obstacles; and cross-latitude flow such as navigating at the interface of water layers with different temperatures or salinities (where the buoyancy and resistance experienced by the robot will change abruptly).
[0023] In a preferred embodiment of the present invention, considering that the road condition data should be able to reflect the amphibious robot's movement posture and movement environment during the movement process, the road condition data includes at least all motion sensor data and all environmental sensor data during the amphibious robot's movement process.
[0024] The motion sensing data includes at least IMU data (such as triaxial angular velocity, triaxial linear acceleration, attitude angle, attitude angular velocity), speed data (such as wheel speed, track speed, underwater propeller speed), motor load data (such as real-time speed and torque of the motors of the propeller, servo motor and track, battery status), and mechanical status data (such as joint angle, track tension, wheel-ground contact force, etc.). Environmental sensing data includes at least aquatic environmental data (such as water flow velocity, water flow turbulence, water depth, water temperature, water quality, and wave data), terrestrial environmental data (ground slope, ground hardness, ground friction, and distance to obstacles), meteorological data (such as wind speed, wind direction, air temperature, air pressure, and humidity), and media characteristic data (such as media density, viscosity, and buoyancy).
[0025] It should be noted that the collection of the aforementioned motion sensing data and environmental sensing data is an existing technological means and will not be elaborated further; implementers may also add or adjust motion sensing data and environmental sensing data according to the configuration of the amphibious robot.
[0026] Since amphibious robots are typically controlled by PID controllers, the data scale of PID control parameters is relatively large. To facilitate subsequent analysis, the PID control parameters can be divided into intervals to summarize several typical models and determine the corresponding control parameter combinations. This can reduce the PID control rules and improve the efficiency of motion optimization.
[0027] Considering that the proportional parameter KP in PID control affects the response speed of the control system, and the response speed is related to the motion environment, for example, a medium KP is needed on a flat land surface to ensure stability, while a larger KP is needed to resist water flow disturbance in turbulent water, and a smaller KP is needed at the junction of land and water to prevent violent oscillations; therefore, the proportional parameter varies greatly under different motion scenarios, and it is necessary to divide the range based on the motion scenario. The integral parameter KI is used to eliminate steady-state error, but it is prone to integral saturation. For example, in noisy and highly disturbed environments (such as turbulent, windy, or obstructed areas), strong integration can improve the system's anti-interference ability, but it will accumulate noise and may lead to severe system oscillations. In relatively stable environments with high accuracy requirements (such as calm waters or land surfaces), strong integration is usually required to ensure accuracy. In such cases, interval division can be directly based on the risk of integral saturation. The differential parameter KD is used to suppress system fluctuations. The larger the differential parameter, the stronger the ability to suppress fluctuations, but it may amplify measurement noise. For example, in motion scenarios with low environmental interference and high control accuracy requirements, strong differential can quickly suppress system oscillations, while in motion scenarios with high environmental interference and high sensor noise levels, weak differential can ensure the stability of the control system. Therefore, the interval can be divided according to the risk of noise amplification. Based on this, in a preferred embodiment of the present invention, the method for obtaining the combination of control parameters includes: The proportional, integral, and derivative parameters of the PID controller of the amphibious robot are used as control parameters. Based on the numerical distribution of the proportional parameters at all historical motion moments, a preset number of initial distribution intervals for the proportional parameters are defined; the preset number is the same as the number of motion scenes; the midpoint of each initial distribution interval is used as the cluster center to cluster all proportional parameters, and the distribution interval of the proportional parameters is determined based on the clustering results. Based on the numerical distribution of integral parameters at all historical motion moments, high-risk and low-risk distribution intervals for integral saturation are defined; based on the numerical distribution of differential parameters at all historical motion moments, high-risk and low-risk distribution intervals for noise amplification are defined. At each historical moment of motion, the combination of control parameters is determined based on the distribution range of each control parameter.
[0028] It should be noted that before dividing the interval, each control parameter needs to be normalized for subsequent division. For example, for the proportional parameter, it should be normalized to the maximum and minimum across all proportional parameters. Implementers can also use function mapping to normalize, such as mapping to the sigmoid function to adjust the range. These are existing technologies and will not be elaborated further.
[0029] Specifically, for the proportional parameters, firstly, based on the numerical distribution of the proportional parameters at all motion moments, an overall distribution interval (i.e., the interval corresponding to the extreme values) is determined; then, the overall distribution interval is divided into a preset number of initial distribution intervals, and the midpoint of each initial distribution interval is used as the cluster center to perform K-means clustering to obtain k clusters; the distribution intervals corresponding to all proportional parameters in each cluster are the divided intervals, such as P1, P2, P3...Pk; It should be noted that the preset number is the same as the number of motion scenes, and is also the preset k value in the K-means clustering algorithm, to ensure that the division of the proportional parameter is highly correlated with the motion scenes; the determination of K-means clustering and distribution intervals are well-known technical means, and will not be elaborated further. For the integral parameter, it can be manually divided directly. First, based on the numerical distribution of the proportional parameter at all motion moments, determine an overall distribution interval, such as [0.1, 0.6], and divide it into a high-risk distribution interval and a low-risk distribution interval for integral saturation, which correspond to the low-risk distribution interval I1 [0.1, 0.3) and the high-risk distribution interval I2 [0.3, 0.6], respectively. For differential parameters, manual division can also be performed directly. First, based on the numerical distribution of the proportional parameters at all motion moments, determine an overall distribution interval, such as [0.2, 1.0], and divide it into a high-risk distribution interval and a low-risk distribution interval for noise amplification, which correspond to the low-risk distribution interval D1 [0.2, 0.5) and the high-risk distribution interval D2 [0.5, 1.0], respectively. Ultimately, the combination of control parameters can be determined based on the distribution range of each control parameter at each historical motion moment; for example, the set of control parameters corresponding to a certain historical motion moment is (P1, I2, D1).
[0030] It should be noted that the acquisition of the above-mentioned PID control parameters is an existing technical means, and the specific acquisition process will not be described in detail. Implementers may also use other clustering algorithms, or adjust the interval division method of the differential and integral parameters according to the actual application situation, that is, adjust the interval endpoint values.
[0031] Considering that the dynamic error of an amphibious robot represents the error generated during the transition from the initial state to the steady state, the steady-state convergence time represents the time required for the transition from the initial state to the steady state, and the steady-state error represents the deviation between the actual value and the target value after the system reaches a steady state; Among them, the larger the dynamic error and the longer the steady-state convergence time, the worse the system's adaptability to large fluctuations in the overall environment; the larger the steady-state error, the lower the system's adaptability to small steady-state fluctuations in the environment; these three factors not only measure the matching degree between PID parameters and environmental parameters, but also reflect the control effect from another perspective. Based on this, in a preferred embodiment of the present invention, the method for obtaining the control effect parameters includes: At each historical moment of motion, the dynamic error, steady-state error, and steady-state convergence time of the amphibious robot are acquired. Based on preset weights, the negative correlation normalization results of the dynamic error, the steady-state error, and the steady-state convergence time are weighted respectively, and the weighted sum is used as the control effect parameter.
[0032] Specifically, the dynamic error, steady-state error, and steady-state convergence time at each historical moment are mapped to the exponential function exp(-x) with the natural constant e as the base for negative correlation normalization. Then, the corresponding negative correlation normalization results are weighted and summed with preset weights of 0.4, 0.2, and 0.4 respectively to obtain the control effect parameters.
[0033] It should be noted that the acquisition of dynamic error, steady-state error, and steady-state convergence time is already existing technology, and the specific acquisition process will not be described in detail. Implementers can also use other negative correlation normalization methods, such as first linear normalization and then subtracting the normalization result by a constant 1. Implementers can also adjust the weights themselves.
[0034] Step S2: Based on the distribution characteristics of control parameter combinations at all historical motion times and the control effect parameters corresponding to each control parameter combination, select all reference control parameter combinations from all control parameter combinations; and obtain the distribution reference range of road condition data according to the road condition data and control effect parameters corresponding to all historical motion times for each reference control parameter combination.
[0035] Considering that the frequently occurring combinations of control parameters during the historical movement of amphibious robots can provide a certain reference for control strategies, and that control effect parameters can characterize the effectiveness of control strategies (combinations of control parameters), screening frequently occurring and effective historical control strategies can provide a reliable reference for subsequent motion optimization. Based on this, the embodiments of the present invention will select all reference control parameter combinations from all control parameter combinations based on the distribution characteristics of control parameter combinations at all historical motion moments and the control effect parameters corresponding to each control parameter combination; the reference control parameter combinations are the frequently occurring and effective historical control strategies.
[0036] Preferably, in one embodiment of the present invention, the method for obtaining the reference control parameter combination includes: Please see Figure 2 The flowchart illustrates a method for obtaining a combination of reference control parameters according to an embodiment of the present invention, specifically including: Step S201: Construct an FP tree based on the control parameter combinations of all historical motion moments. The nodes in the FP tree are individual combination parameters in the control parameter combinations.
[0037] Given that FP-Growth (Frequent Pattern Growth) is an efficient frequent itemset mining algorithm, often used in association rule learning in data mining; it compresses data by constructing an FP-tree and recursively mines frequent itemsets, showing high efficiency when processing large datasets, and is especially suitable for large-scale data mining tasks, based on this, frequent and effective combinations of control parameters can be mined by constructing an FP-tree.
[0038] It should be noted that the construction of FP-trees is a well-known technique and will not be elaborated further; assuming the set of control parameters is (P1, I2, D1), then P1 is one of the individual combination parameters.
[0039] Step S202: Based on the control effect parameters and frequency of occurrence of each control parameter combination, obtain the reference index of each individual combination parameter in the FP tree.
[0040] Since it is necessary to select frequently occurring and effective historical control strategies as reference strategies, that is, to select all reference control parameter combinations from all control parameter combinations, we first evaluate the combination reference index that reflects the reference value of the strategy based on the control effect parameters and the frequency of occurrence, in order to prepare for subsequent FP-tree pruning and rapid extraction of reference control parameter combinations.
[0041] Preferably, in one embodiment of the present invention, considering that each control parameter combination corresponds to a frequency of occurrence, and the higher the frequency of occurrence, the more frequently the strategy is applied, the maximum frequency of occurrence among all control parameter combinations is used as a comparative reference to evaluate the relative frequency of each control parameter combination; furthermore, considering that the combination frequency parameter also indirectly characterizes the relative frequency of each individual combination parameter in the control strategy, each individual combination parameter corresponds to a relative frequency in its corresponding control parameter combination, and thus the control effect parameter of each control parameter combination can be comprehensively evaluated; therefore, the method for obtaining the reference index includes: The ratio of the frequency of occurrence of each control parameter combination to the maximum frequency of occurrence among all control parameter combinations is taken as the combination frequency parameter. Under each control parameter combination, for each individual combination parameter in the FP tree, the control effect parameter of each control parameter combination in which it is located is used to weight and sum the corresponding combination frequency parameter to obtain the sub-reference coefficient. The reference index is obtained by combining the sub-reference coefficients in all control parameter combinations.
[0042] Specifically, the frequency of occurrence of each control parameter combination is used as the numerator, the maximum frequency of occurrence is used as the denominator, and the ratio of the fractions is used as the combination frequency parameter. Under each control parameter combination, for each individual combination parameter in the FP tree, the control effect of each control parameter combination in which it is located is multiplied by the corresponding combination frequency parameter, and the product is used as the sub-reference coefficient of the individual combination parameter under that control parameter combination. Then, the sum of the sub-reference coefficients under all control parameter combinations is used as the reference index of each individual combination parameter.
[0043] Step S203: Based on the reference index and the preset minimum support, FP tree is pruned, and all corresponding control parameter combinations in the pruned FP tree are used as reference control parameter combinations.
[0044] As an example, the combined reference index is set to 0.8 and the preset minimum support is set to 0.2, which implementers can adjust themselves; then the FP tree is pruned (a well-known technique, which will not be described in detail); after pruning, all the corresponding control parameter combinations can be obtained (a well-known technique, which will not be described in detail), and these control parameter combinations are used as reference control parameter combinations.
[0045] In another embodiment of the present invention, the ratio of the frequency of occurrence of each control parameter combination to the maximum frequency of occurrence among all control parameter combinations can be directly used as the combination frequency parameter; the combination frequency parameter is weighted by the control effect parameter to obtain the combination reference index of each control parameter combination; then, based on a set threshold such as 0.8, control parameter combinations with a combination reference index greater than the threshold are selected as reference control parameter combinations.
[0046] After filtering out valuable historical control strategies (reference control parameter combinations) from all motion sample information, the corresponding road condition data in the historical motion process can be further determined. Thus, given a road condition data, effective motion control can be carried out directly by using an effective historical control strategy.
[0047] Based on this, the distribution reference range of road condition data is obtained according to the road condition data and control effect parameters at all historical motion moments corresponding to each combination of reference control parameters. The distribution reference range of road condition data reflects the historical environment or motion information corresponding to each effective historical control strategy, providing a basis for comparison for matching appropriate control strategies in the future.
[0048] Preferably, in one embodiment of the present invention, considering the road condition data and their differences at all different historical motion times corresponding to each combination of reference control parameters can help establish the distribution of physical characteristics of the environment, thereby understanding the road condition distribution; while the control effect parameters reflect the quality or importance of the data collected at that historical motion time, thus enabling better mining of similar road condition information; the method for obtaining the distribution reference range includes: Using the control effect parameters at each historical motion moment as weights, and based on the differences between road condition data at different historical motion moments, weighted mean drift clustering is performed on the road condition data at all historical motion moments corresponding to each combination of reference control parameters. The distribution reference range of the road condition data is determined based on the clustering results.
[0049] It should be noted that weighted mean drift clustering is a well-known technique and will not be elaborated further; each cluster corresponds to a distribution reference range, that is, the range of values corresponding to all road condition data in the cluster corresponds to a distribution reference range.
[0050] Step S3: At the current moment of motion, determine the control parameters of the amphibious robot and perform motion control based on the distribution reference range of the current road condition data of the amphibious robot.
[0051] The analysis in steps S1-S2 above can be regarded as the construction of a historical control strategy library, which can be applied to the current motion control of the amphibious robot. After determining the distribution reference range of road condition data under each combination of reference control parameters, the historical reference provided for the current road condition data of the amphibious robot can be further determined to determine its distribution reference range, thereby evaluating its motion scenario and recommended control strategy, and then determining the control parameters of the amphibious robot and performing motion control on it.
[0052] Preferably, in one embodiment of the present invention, the method for determining the control parameters of an amphibious robot includes: The initial control parameter is obtained by taking the midpoint of the distribution interval of each control parameter in the reference control parameter combination corresponding to the current road condition data distribution reference range; the initial control parameter is then fine-tuned based on the preset fuzzy rule base to obtain the control parameter.
[0053] Specifically, based on the distribution reference range of the current road condition data, a corresponding combination of reference control parameters is determined, and the midpoint of the distribution interval of each control parameter in the combination of reference control parameters is used as the initial control parameters (PID control parameters) of the amphibious robot at the current moment of motion. Then, at the current moment of motion, the real-time error e and the error change rate ec are calculated. The real-time error e and the error change rate ec are converted into fuzzy language values according to a preset fuzzy rule library for fine-tuning, thereby obtaining the control parameters for real-time motion control of the amphibious robot.
[0054] Please see Figure 3 The diagram shows an example of a preset fuzzy rule base provided by the present invention. Implementers can also adjust the fuzzy rule base according to actual applications.
[0055] It should be noted that the calculation of real-time error e and error change rate ec, as well as the subsequent fuzzy control, are well-known technologies and will not be elaborated further; the adjustment of the above control parameters must not exceed the range corresponding to the reference control parameter combination.
[0056] Based on the same inventive concept, the present invention also proposes a motion optimization system for an amphibious robot, which includes a historical data acquisition module 101, a historical data analysis module 102, and a motion analysis and control module 103.
[0057] Historical data acquisition module 101: used to acquire motion sample information of the amphibious robot in each motion scenario. The motion sample information includes road condition data, control parameter combinations and control effect parameters at each historical motion moment. The motion scenarios include various land scenarios, water scenarios and water-land transition scenarios. Historical data analysis module 102: Based on the distribution characteristics of control parameter combinations at all historical motion times and the corresponding control effect parameters for each control parameter combination, it filters out all reference control parameter combinations from all control parameter combinations; and obtains the distribution reference range of road condition data according to the road condition data and control effect parameters corresponding to each reference control parameter combination at all historical motion times. Motion analysis and control module 103: used to determine the control parameters of the amphibious robot and perform motion control based on the distribution reference range of the current road condition data of the amphibious robot at the current moment of motion.
[0058] It should be noted that each module is used to perform the content described in steps S1-S3 above, and will not be repeated here.
[0059] Based on the same inventive concept, the present invention also proposes a motion optimization device for an amphibious robot. The device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the motion optimization method for an amphibious robot described in steps S1-S3 above.
[0060] In summary, this invention first acquires motion sample information of the amphibious robot in each motion scenario. This motion sample information includes road condition data, control parameter combinations, and control effect parameters for each historical motion moment. Then, based on the distribution characteristics of the control parameter combinations across all historical motion moments and the corresponding control effect parameters for each combination, all reference control parameter combinations are selected. According to the road condition data and control effect parameters corresponding to each reference control parameter combination across all historical motion moments, a distribution reference range for the road condition data is obtained. At the current motion moment, based on the distribution reference range of the amphibious robot's current road condition data, the control parameters of the amphibious robot are determined, and its motion is controlled. This invention, based on the effective reference provided by the amphibious robot's historical motion data and control strategies, can adaptively match control strategies based on the current motion situation, avoiding rule conflicts and improving the motion control effect of the amphibious robot.
[0061] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0062] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.
Claims
1. A motion optimization method for an amphibious robot, characterized in that, The method includes: Acquire motion sample information of the amphibious robot in each motion scenario. The motion sample information includes road condition data, control parameter combinations and control effect parameters at each historical motion moment. The motion scenarios include various land scenarios, water scenarios and water-land transition scenarios. Based on the distribution characteristics of control parameter combinations at all historical motion times, and the corresponding control effect parameters for each control parameter combination, all reference control parameter combinations are selected from all control parameter combinations; based on the road condition data and control effect parameters corresponding to each reference control parameter combination at all historical motion times, the distribution reference range of road condition data is obtained. At the current moment of motion, based on the distribution reference range of the current road condition data of the amphibious robot, the control parameters of the amphibious robot are determined and its motion is controlled.
2. The motion optimization method for an amphibious robot according to claim 1, characterized in that, The road condition data includes at least all motion sensor data and all environmental sensor data during the movement of the amphibious robot.
3. The motion optimization method for an amphibious robot according to claim 1, characterized in that, The method for obtaining the control parameter combination includes: The proportional, integral, and derivative parameters of the PID controller of the amphibious robot are used as control parameters. Based on the numerical distribution of the proportional parameters at all historical motion moments, a preset number of initial distribution intervals for the proportional parameters are defined; the preset number is the same as the number of motion scenes; the midpoint of each initial distribution interval is used as the cluster center to cluster all proportional parameters, and the distribution interval of the proportional parameters is determined based on the clustering results. Based on the numerical distribution of integral parameters at all historical motion moments, high-risk and low-risk distribution intervals for integral saturation are defined; based on the numerical distribution of differential parameters at all historical motion moments, high-risk and low-risk distribution intervals for noise amplification are defined. At each historical moment of motion, the combination of control parameters is determined based on the distribution range of each control parameter.
4. The motion optimization method for an amphibious robot according to claim 1, characterized in that, The method for obtaining the control effect parameters includes: At each historical moment of motion, the dynamic error, steady-state error, and steady-state convergence time of the amphibious robot are acquired. Based on preset weights, the negative correlation normalization results of the dynamic error, the steady-state error, and the steady-state convergence time are weighted respectively, and the weighted sum is used as the control effect parameter.
5. The motion optimization method for an amphibious robot according to claim 1, characterized in that, The method for obtaining the reference control parameter combination includes: An FP tree is constructed based on the combination of control parameters at historical motion moments, where each node in the FP tree is a single combination parameter of the control parameter combination. Based on the control effect parameters and frequency of occurrence of each control parameter combination, obtain the reference index of each individual combination parameter in the FP tree; Based on the reference index and the preset minimum support, the FP tree is pruned, and all corresponding control parameter combinations in the pruned FP tree are used as reference control parameter combinations.
6. The motion optimization method for an amphibious robot according to claim 5, characterized in that, The methods for obtaining the reference index include: The ratio of the frequency of occurrence of each combination of control parameters to the maximum frequency of occurrence among all combinations of control parameters is used as the combination frequency parameter. Under each combination of control parameters, for each individual combination parameter in the FP tree, the control effect parameter of each combination of control parameters in which it is located is used to weight and sum the corresponding combination of frequent parameters to obtain the sub-reference coefficient; The reference index is obtained by combining the sub-reference coefficients under all combinations of control parameters.
7. The motion optimization method for an amphibious robot according to claim 1, characterized in that, The method for obtaining the distribution reference range includes: Using the control effect parameters at each historical motion moment as weights, and based on the differences between road condition data at different historical motion moments, weighted mean drift clustering is performed on the road condition data at all historical motion moments corresponding to each combination of reference control parameters, and the distribution reference range of the road condition data is determined based on the clustering results.
8. The motion optimization method for an amphibious robot according to claim 3, characterized in that, Methods for determining the control parameters of amphibious robots include: The initial control parameters are determined by combining the reference control parameters corresponding to the distribution reference range of the current road condition data; the initial control parameters are then fine-tuned based on a preset fuzzy rule base to obtain the final control parameters.
9. A motion optimization system for an amphibious robot, characterized in that, The system includes: Historical data acquisition module: used to acquire motion sample information of amphibious robot in each motion scenario. The motion sample information includes road condition data, control parameter combinations and control effect parameters at each historical motion moment. The motion scenarios include various land scenarios, water scenarios and water-land transition scenarios. Historical data analysis module: Based on the distribution characteristics of control parameter combinations at all historical motion times, and the corresponding control effect parameters for each control parameter combination, it filters out all reference control parameter combinations from all control parameter combinations; and obtains the distribution reference range of road condition data based on the road condition data and control effect parameters corresponding to each reference control parameter combination at all historical motion times. Motion analysis and control module: used to determine the control parameters and perform motion control on the amphibious robot based on the distribution reference range of the current road condition data at the current moment of motion.
10. A motion optimization device for an amphibious robot, the device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the motion optimization method for an amphibious robot as described in any one of claims 1 to 8.