Multi-vehicle type hook unhooking robot operation mode switching method and device

Abstract

The application provides a multi-vehicle-type uncoupling robot operation mode switching method and device, relates to the vehicle coupler uncoupling technical field, and comprises the following steps: carrying out state sensing on a target vehicle to obtain the vehicle type and coupler state vector of the target vehicle; based on the vehicle type and the coupler state vector of the target vehicle, combining the structure prior of the current operation scene, generating a plurality of coupler candidate operation states; performing a trial action on the coupler of the target vehicle by the uncoupling robot, and acquiring trial interaction feedback data in the trial process; based on the trial interaction feedback data, matching and evaluating each coupler candidate operation state to obtain a matching credibility; switching the operation mode of the uncoupling robot based on the matching credibility, and performing the uncoupling operation based on the determined operation mode. The application solves the problem that it is difficult to reliably determine the real operation state of the coupler in the existing uncoupling operation only by relying on the appearance or geometric information.

Description

Technical Field

[0001] This invention relates to the field of coupler and uncoupling technology, and more specifically, to a method and apparatus for switching the operation mode of a multi-vehicle coupler and uncoupling robot. Background Technology

[0002] In railway freight marshalling, vehicle maintenance, and automated shunting operations, vehicles are typically mechanically connected via couplers. Uncoupling is a crucial step in the train marshalling and vehicle separation process. With the increasing automation and unmanned operation of railways, utilizing uncoupling robots to replace manual labor in uncoupling operations has become an important development trend.

[0003] However, the existing railway vehicles have evolved over a long period into a practical application pattern where multiple models and couplers coexist. Couplers between different models vary significantly in structural form, opening method, spatial position, and stress characteristics. Even for the same model, couplers may exhibit different operating states in actual operation due to factors such as wear, deformation, foreign object jamming, or partial opening. This makes it difficult for uncoupling robots to reliably uncouple based solely on a pre-set single operating strategy during actual operations.

[0004] In existing technologies, the operation mode of uncoupling robots is mostly based on the assumption of fixed vehicle models or predefined action procedures. They typically rely on manual configuration of vehicle model information or simple sensors to determine whether the coupler is in a standard state. If the actual coupler state is inconsistent with the preset model, it can easily lead to uncoupling failure, abnormal force, or even equipment damage. At the same time, relying solely on visual perception or position detection is difficult to accurately distinguish the degree of coupler opening, restricted state, and hidden structural features, resulting in incomplete perception and a high probability of misjudgment in complex operating environments. Summary of the Invention

[0005] The purpose of this invention is to provide a method, apparatus, device, and readable storage medium for switching operating modes of a multi-vehicle unhooking robot, in order to improve the aforementioned problems. To achieve the above objective, the technical solution adopted by this invention is as follows:

[0006] Firstly, this application provides a method for switching operating modes of a multi-vehicle unhooking robot, including:

[0007] The target vehicle is state-aware to obtain the vehicle type and coupler state vector. The coupler state vector includes the appearance features, spatial pose information and visible structure information of the coupler.

[0008] Based on the vehicle type and the coupler state vector of the target vehicle, and combined with the structural prior of the current operation scenario, multiple candidate coupler operation states are generated.

[0009] The unhooking robot performs a probing action on the hook of the target vehicle to obtain the probing interaction feedback data during the probing process;

[0010] Based on the trial interaction feedback data, the matching and evaluation of each coupler candidate operation status is performed to obtain the matching credibility.

[0011] The operation mode of the unhooking robot is switched based on the matching confidence level, and the unhooking operation is performed based on the determined operation mode.

[0012] Secondly, this application also provides a multi-vehicle unhooking robot operation mode switching device, including:

[0013] The perception module is used to perceive the state of the target vehicle and obtain the vehicle type and coupler state vector. The coupler state vector includes the appearance features, spatial pose information and visible structure information of the coupler.

[0014] The generation module is used to generate multiple candidate coupler operation states based on the target vehicle model and the coupler state vector, combined with the structural prior of the current operation scenario.

[0015] The probing module is used to perform probing actions on the hook of the target vehicle through the unhooking robot and obtain probing interaction feedback data during the probing process;

[0016] The matching module is used to perform matching evaluation on each of the candidate coupler operation states based on the trial interaction feedback data, and obtain the matching credibility.

[0017] The switching module is used to switch the operation mode of the unhooking robot based on the matching confidence level, and to perform the unhooking operation based on the determined operation mode.

[0018] The beneficial effects of this invention are as follows:

[0019] (1) This invention combines visual perception with physical probing. In response to the situation of diverse coupler models, continuous changes in opening degree, and complex restricted or abnormal release states under multiple vehicle conditions, a unified and restricted probing action is introduced on the basis of generating multiple coupler candidate operating states. By collecting probing interaction feedback data such as displacement, force and damping, dynamic information reflecting the real mechanical characteristics of the coupler internal mechanism is obtained. This makes multiple candidate operating states that were originally indistinguishable in terms of geometry and visibility distinguishable at the physical interaction level, thereby improving the accuracy and reliability of coupler state recognition.

[0020] (2) This invention also constructs a trial response expectation model and adjusts the distribution of trial response characteristics in conjunction with environmental information, thereby achieving state matching evaluation under different environmental conditions such as temperature, humidity, and light, and avoiding misjudgment of operation decisions caused by environmental changes. On this basis, the operation mode of the uncoupling robot is adaptively switched based on the matching credibility, enabling the robot to dynamically adjust its motion trajectory, speed planning, and force control parameters according to different actual states of the coupler, significantly reducing the risk of misoperation while ensuring the success rate of the operation. Compared with fixed process or single mode uncoupling methods, it has stronger versatility and scalability, and can adapt to multiple vehicle types, multiple states, and complex working conditions.

[0021] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of the multi-vehicle unhooking robot operation mode switching method described in this embodiment of the invention;

[0024] Figure 2 This is a schematic diagram of the multi-vehicle unhooking robot operation mode switching device described in an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0026] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0027] Example 1:

[0028] This embodiment provides a method for switching the operation mode of a multi-vehicle unhooking robot.

[0029] It is understood that this invention is applicable to multi-vehicle uncoupling operations within the same railway operation scenario, such as railway tippler uncoupling stations, fixed uncoupling operation stations in marshalling yards, or other railway operation areas with fixed uncoupling requirements. In such scenarios, although different types of freight vehicles (such as open wagons, boxcars, bulk freight trains, etc.) differ in car body structure, dimensional parameters, and loading methods, they are usually connected using coupler structures that conform to unified railway technical standards, such as No. 13 coupler or No. 17 coupler.

[0030] In the aforementioned application scenarios, the uncoupling robot is deployed in a fixed manner in a designated work area. Once the train enters the predetermined work position, the robot performs status perception and probing interaction operations on the couplers of the target vehicle. Simultaneously, the uncoupling robot can select and switch between multiple candidate work modes to adapt to the uncoupling requirements of different train models and different coupler opening, restricted, or abnormal states. This enables the uncoupling robot to complete automatic uncoupling operations for multiple train models within the same work scenario without changing or frequently adjusting the robot's hardware structure, improving the versatility, reliability, and efficiency of uncoupling operations. It is particularly suitable for railway automation environments with mixed train models and high uncertainty in coupler status.

[0031] See Figure 1 The figure shows that the method includes steps S1, S2, S3, S4 and S5.

[0032] Step S1: Perform state perception on the target vehicle to obtain the vehicle type and coupler state vector. The coupler state vector includes the appearance features, spatial pose information and visible structure information of the coupler.

[0033] In railway operation environments with mixed train models and uncertain coupler status, structured state perception of the target cars and their couplers is performed. The originally vague and incomplete perception results are transformed into coupler state vectors containing appearance, spatial and structural information, providing reliable input for subsequent generation of coupler candidate operation states and operation mode switching.

[0034] Step S1 includes:

[0035] Step S11: Collect the location information of the target vehicle within the work area and determine the uncoupling end of the target vehicle;

[0036] In applications such as railway tippers and fixed uncoupling stations, after a train enters the work area, its uncoupling end is usually located within a predefined position range of the work area. This step identifies and matches the overall position of the target vehicle to determine the end of the vehicle that needs to be uncoupled, providing a spatial starting point for subsequent coupler sensing.

[0037] Step S12: Obtain the vehicle model based on the explicit identification information of the target vehicle or the prior information of the operating system;

[0038] In this step, explicit identification information may include vehicle number, vehicle type code, electronic tag, or external identification mark. Prior information for the operating system may come from the dispatching system, work plan, or historical grouping information.

[0039] Although vehicle model information does not directly determine the uncoupling action, by introducing vehicle model information in the early stages of perception, we can avoid including the coupler states that are impossible for different vehicle models in subsequent judgments.

[0040] Step S13: Based on the vehicle model of the target vehicle, obtain the prior spatial position of the corresponding coupler;

[0041] In this step, the coupler spatial position prior is used to describe the theoretical positional distribution range of the coupler relative to the vehicle end under the conditions of this vehicle model, such as the height range, longitudinal position range, and allowable attitude deviation range. By introducing the coupler spatial position prior, the coupler search space can be limited to a reasonable range of the vehicle model, avoiding misidentification of buffer devices, vehicle body edges, etc., as couplers.

[0042] Step S14: Under the prior knowledge of the coupler's spatial position, the coupler of the target vehicle is perceived based on the uncoupling operation end to obtain the coupler state vector.

[0043] Step S14 includes:

[0044] Step S141: At the end of the uncoupling operation, based on the prior spatial position of the coupler, construct the coupler sensing area in the working coordinate system of the uncoupling robot;

[0045] In this step, the coupler sensing area is a local sensing space obtained by further trimming from the spatial location prior, which is used to centrally process sensing data related to the coupler.

[0046] Step S142: Within the coupler sensing area, sample the appearance contour of the coupler and extract its appearance features;

[0047] In this step, within the coupler sensing area, image segmentation of the coupler is performed based on visual images. Specifically, coupler images under the coupler sensing area are acquired, and key coupler structures such as the coupler body, coupler opening structure, and locking structure are obtained through a trained image segmentation model as appearance features (including the category label, contour region, or pixel mask of each key coupler structure in the image).

[0048] Step S143: Extract the key structure of the coupler based on appearance features, and spatially locate the key structure of the coupler to obtain spatial pose information. The spatial pose information includes the position coordinates of the key points of the coupler in the working coordinate system, the main axis direction and normal direction of the coupler, and the relative distance and relative height of the coupler relative to the uncoupling operation end.

[0049] In this step, after obtaining the key structure of the coupler based on image recognition, the corresponding three-dimensional point cloud data is collected to spatially locate the key structure of the coupler and obtain the spatial pose information of the coupler.

[0050] Specifically, the two-dimensional regions of key coupler structures in the image are mapped to corresponding three-dimensional point cloud subsets. Noise reduction and clustering are performed on the point clouds corresponding to each key coupler structure, and their geometric center points are calculated as the coupler key points. Then, PCA principal component analysis is performed on the point cloud corresponding to the coupler head body, with the direction of the largest eigenvalue taken as the coupler principal axis direction. In a plane perpendicular to the coupler principal axis direction, the coupler normal direction is determined by the point cloud thickness or distribution density. Based on the positional relationship between the coupler key points and the uncoupling operation end, the relative distance and relative height are calculated.

[0051] Step S144: Based on the key structure and spatial pose information of the coupler, and combined with the prior knowledge of the coupler composition, identify the visible structural information of the coupler in the current state.

[0052] In this step, the coupler composition prior is used to describe the structural composition relationship of the standard coupler and the relative positional distribution characteristics of each component structure in the coupler's local coordinate system. Based on the coupler composition prior, the visibility of each identified key coupler structure is determined within its corresponding theoretical location range: two-dimensional visibility is determined based on its contour integrity and occlusion ratio in the image, and three-dimensional exposure is determined based on the number of points and spatial continuity of its corresponding point cloud. When the number of points in the point cloud of a key structure within its theoretical location range is lower than a preset threshold, or when it is severely occluded in the image, the structure is determined to be invisible or limitedly visible in the current state, and finally, the visible structural information of the coupler in the current state is constructed.

[0053] Step S2: Based on the target vehicle's model and the coupler state vector, and combined with the structural priors of the current operating scenario, generate multiple coupler candidate operating states;

[0054] In step S2, generating multiple candidate coupler operation states includes:

[0055] Step S21: Based on the vehicle type and the current operating scenario, construct the coupler operation state space, which includes the coupler types allowed for the vehicle type, coupler opening degree parameters, and coupler restricted status indicators.

[0056] Step S22: Filter coupler configuration categories that match the appearance features in the coupler operation state space to obtain a candidate set of coupler configurations;

[0057] In this step, the pixel area and aspect ratio of each key structure of the coupler are calculated using appearance features to obtain single-structure scale features. Then, the relative proportion features between structures are calculated using these single-structure scale features, such as the area ratio of the hook opening to the hook head body. Additionally, the centroid of the pixel mask for each key structure of the coupler is calculated to obtain the spatial layout features between structures. These single-structure scale features, relative proportion features between structures, and spatial layout features between structures are used as the target configuration features.

[0058] For the coupler model in the coupler operation state space, prior configuration features obtained from multiple angles are recorded. The similarity between the target configuration features and the prior configuration features is calculated, and coupler models with a similarity greater than a preset similarity threshold are used as the configuration candidate set.

[0059] Step S23: Perform geometric consistency constraints based on the spatial pose information of the key structure of the coupler, and derive the feasible range of the coupler opening degree parameters;

[0060] It is understandable that, for the same coupler model, the opening process is essentially a rigid body mechanism motion process with restricted rotation. The hook tongue rotates relative to the hook head body around a fixed axis, and the overall geometric topology of the coupler remains unchanged during operation. Therefore, neglecting elastic deformation and extreme jamming, the position and attitude of the key structure of the coupler in space can be determined by the coupler opening degree parameter. Sure.

[0061] In the uncoupling operation station, the position of the uncoupling robot's working end in the working coordinate system is basically fixed. Therefore, the relative distance, relative height, and orientation information of the key structure of the coupler relative to the uncoupling working end can all be expressed as parameters related to the opening. The functional relationship is as follows. Within the normal operating range of the coupler, the functional relationship exhibits a continuous and stable trend of change, and some spatial pose quantities show monotonic or quasi-monotonic variation characteristics within the effective operating range.

[0062] Therefore, based on physical and geometric characteristics, it is possible to pre-build using offline modeling methods. The mapping relationship between the coupler and the spatial pose information, and the use of the spatial pose information obtained by sensing during actual operation. By applying reverse constraints, a feasible range of activation parameters that conform to the current operation status can be derived.

[0063] Specifically, a three-dimensional geometric model of the coupler and its connecting mechanism is established for each coupler model, and the coupler opening parameters are defined. (This embodiment uses the angle parameter of the hook tongue relative to the hook head body) Discrete sampling is performed within the range, for each discrete... Through multibody kinematics or finite element analysis, considering the geometric constraints and limiting conditions of the mechanism, the corresponding coupler spatial pose information is calculated and obtained, and then... It is associated with and stored in conjunction with the corresponding spatial pose information. Indicates the minimum opening value. This indicates the maximum open value.

[0064] Based on the above offline calculation results, establish The mapping relationship between spatial pose information and spatial position information can be represented by multinomial regression or neural network models.

[0065] During the operation, based on the spatial pose information of the target vehicle, a preset error tolerance threshold is introduced to apply reverse constraints to the above mapping relationship, and vehicles that meet the consistency conditions with the current perception results in geometric features such as principal axis direction, normal direction, relative distance, and relative height are selected. The set of values ​​is used to deduce the feasible range of coupler opening degree parameters that meet the current operating conditions.

[0066] Step S24: Based on the visible structural information, identify whether the coupler has any structural features that restrict or abnormally release, and determine the candidate set of status indicators;

[0067] In this step, the coupler opening is achieved through the restricted rigid body rotation and mechanism linkage. During normal opening, the obstruction, visibility, and relative position relationships between the hook tongue, hook opening, and locking element satisfy the design principles. However, if the following situations occur, such as spring failure, incomplete disengagement of the locking element, or the hook tongue being jammed by a foreign object, even... It may be geometrically feasible, but its structural visibility will be off.

[0068] Specifically, based on visible structural information and target configuration features, visibility features are obtained, including visibility status indicators (whether it is detected), visible area ratio (current visible area / theoretical maximum visible area) and structural boundary integrity indicators (reflecting whether it is partially occluded).

[0069] For the visibility characteristics of multiple key coupler structures, the relative visibility order relationship between the key coupler structures is obtained to determine whether there are abnormal occlusion patterns, such as a large opening angle of the coupler tongue but the locking part is still partially visible, or the coupler opening area should be fully exposed but is occluded. Then, the occlusion relationship between structures is encoded as an occlusion consistency feature.

[0070] Simultaneously, establishing visible prior relationships offline, in different Below, the theoretical visibility state of each coupler's key structure, the range of changes in the visible area ratio, the legal combinations of occlusion relationships, and the corresponding states of restricted or abnormal release.

[0071] During the operation, after acquiring the visibility features and occlusion consistency features of the target vehicle, matching is performed based on the visibility prior relationship. If there is a significant deviation from the prior pattern, it is determined that there are structural features of restricted or abnormal release, thereby obtaining different state identifier candidates.

[0072] Understandably, the physical essence of restricted release is that the mechanism still moves according to the original kinematic chain, but with additional constraints (friction, slight jamming, insufficient elasticity). The mapping relationship between visibility and the physical property still holds, but interval compression or offset occurs, and no illegal combination of occlusion relationships between structures occurs. Abnormal release, however, occurs when the kinematic chain of the mechanism is disrupted or severely deviated. The visibility of the structure can no longer be determined solely by the detection of obstruction relationships that do not conform to the design principles of the coupler mechanism, or the absence of critical structures. For example, illegal obstruction relationships may occur, or the locking components may be in a large position. The hook opening is still completely obscured, or the relative layout of the structure does not meet the rigid body constraints.

[0073] Step S25: Based on the configuration candidate set, the feasible range of the coupler opening degree parameters, and the state identifier candidate set, perform joint constraints and screening on the coupler operation state space to obtain multiple coupler candidate operation states that meet the current sensing conditions.

[0074] In this step, the candidate operation status of the coupler is shown in Table 1.

[0075] Table 1. Schematic diagram of coupler candidate operation status

[0076]

[0077] Step S3: The unhooking robot performs a probing action on the hook of the target vehicle and obtains the probing interaction feedback data during the probing process;

[0078] It is understandable that multiple candidate coupler operating states may be valid geometrically and visually, but their actual mechanical accessibility, release resistance, and presence of jamming cannot be reliably judged visually alone. Therefore, a low-risk, restricted probing action is introduced. Through the physical interaction feedback between the uncoupling robot and the coupler, dynamic information reflecting the actual mechanism state is obtained, thereby enabling accurate state identification and strategy selection.

[0079] Step S3 includes:

[0080] Step S31: Based on the safety constraints of the current work scenario, initialize the trial action parameters, which include the trial movement direction, the upper limit of the trial displacement, and the upper limit of the trial force;

[0081] In this embodiment, uniform trial action parameters are used because different coupler operating states will exhibit differences in displacement response, force variation characteristics, and damping characteristics due to variations in the internal mechanism's degrees of freedom, friction conditions, and constraints. Furthermore, by eliminating the influence of the trial action itself, the trial interaction feedback data has good comparability, thus reflecting the differences in the coupler's actual state.

[0082] Step S32: Control the unhooking robot to perform a probing action on the target vehicle's hook along the probing motion direction, based on the probing motion parameters. The probing action includes a micro-displacement propulsion with a limited stroke and a controlled contact process.

[0083] In this step, the limited-stroke micro-displacement propulsion is used to gradually establish a controlled contact relationship between the end effector of the uncoupling robot and the key structure of the coupler, ensuring that the contact process remains in a quasi-static or low-dynamic range, thereby avoiding inertial interference introduced by high-speed impacts. The controlled contact process can stimulate a real mechanical response in the internal mechanism of the coupler without causing significant changes in the coupler's position.

[0084] Step S33: During the execution of the probing action, probing interaction feedback data is collected through the contact component at the end of the unhooking robot. The probing interaction feedback data includes displacement response data of the end displacement of the unhooking robot changing with time, data on the change of force at the end of the unhooking robot, and data on the change of damping during the contact process between the coupler and the contact component.

[0085] In this step, the same end-effector propulsion command is applied to the uncoupling robot. After the contact component makes physical contact with the coupler, the resistance characteristics of the coupler's internal mechanism are fed back to the contact component through the contact interface. This causes the actual displacement, force changes, and damping characteristics of the contact component to vary depending on the coupler's state. Therefore, by collecting the trial interaction feedback data of the uncoupling robot's end-effector contact component, it is possible to achieve differentiated perception of the coupler's actual mechanical state, making multiple visually indistinguishable candidate coupler operating states distinguishable at the physical interaction level.

[0086] Step S4: Based on the trial interaction feedback data, perform a matching evaluation on each of the candidate coupler operation states to obtain the matching confidence level;

[0087] Step S4 includes:

[0088] Step S41: By extracting features from the trial interaction feedback data, construct an actual trial feature vector including multiple trial response features;

[0089] In this step, the actual trial feature vector is obtained as follows:

[0090] ;

[0091] In the formula, This represents the actual probe feature vector. This represents the maximum actual displacement during the probing process. Indicates the maximum contact force. Indicates equivalent stiffness. Indicates equivalent damping, This indicates the energy dissipation during the probing process. Indicates the force response setup time.

[0092] The formula for calculating the trial response characteristics is as follows:

[0093] ;

[0094] ;

[0095] ;

[0096] In the formula, Indicates equivalent stiffness. This indicates the increase in contact force during the initial contact process. This represents the displacement increment during the trial contact process. Indicates equivalent damping, Indicates the number of sampling points within the interval. This indicates the time interval during the trial action when the end contact component of the unhooking robot achieves stable physical contact with the coupler. Indicates in Contact force at all times Indicates contact parts at The actual speed at that moment This indicates the energy dissipation during the probing process. Indicates the time of first contact Displacement at time, Indicates the end time of the probing action. Displacement at time, Indicates contact force with respect to displacement The function.

[0097] Step S42: Obtain environmental information for the current work scenario;

[0098] In this step, environmental information includes ambient temperature, ambient humidity, and visible light intensity.

[0099] Step S43: For each coupler candidate operating state, combine the environmental information and obtain the candidate trial response feature distribution through the trial response expectation model;

[0100] In this step, under uniform trial action parameters, multiple trial actions are performed under different environmental conditions for different known coupler operating states, and the corresponding trial response feature vectors are collected.

[0101] For the same coupler operating condition, the primary determinant of its trial response characteristics stems from the internal mechanical state of the coupler itself. By establishing the baseline characteristic mean and baseline characteristic covariance under standard environmental conditions, environmental interference can be eliminated to the greatest extent possible, thereby extracting the intrinsic response characteristics that reflect the true mechanical state of the coupler. Specifically, for a fixed state, under standard environmental conditions, the baseline characteristic mean and baseline characteristic covariance of the known coupler operating state are calculated:

[0102] ;

[0103] ;

[0104] In the formula, Indicates the known coupler operation status The baseline characteristic mean, Represents the mathematical expectation operator. Indicates the known coupler operation status The baseline characteristic covariance, Describes the covariance operator. This represents the feature vector of the trial response to a known coupler operating state. This represents the standard environment vector.

[0105] The baseline characteristic covariance can be used to characterize the dispersion of trial responses caused by factors such as manufacturing tolerances, friction fluctuations, and assembly differences under the same operating conditions. The predefined standard environment vector serves as the benchmark reference environment for modeling the trial response characteristics. Its corresponding ambient temperature, ambient humidity, and visible light intensity are within the standard range under normal operating conditions of the coupler.

[0106] Meanwhile, in actual operating scenarios, ambient temperature, humidity and light conditions will affect the friction characteristics, material elasticity and sensor measurement noise of the coupler. However, this effect usually manifests as a translation or extension based on the reference response, without changing the basic mechanical mode of the coupler's operating state.

[0107] Environmental factors can alter the average level of characteristics on the trial response, for example, an increase in temperature can lead to an overall decrease in friction, and they can also introduce additional uncertainties, such as changes in humidity leading to increased response fluctuations. Therefore, by independently modeling the characteristic mean and covariance using the environmental sensitivity coefficient matrix and the environmental uncertainty adjustment matrix, we can more accurately reflect the dual influence mechanism of the environment on the trial response.

[0108] Specifically, for and Environmental adaptive adjustments were performed to obtain the mean and covariance of environmental characteristics:

[0109] ;

[0110] ;

[0111] In the formula, Indicates the known coupler operation status and environmental condition vector The mean environmental characteristics under the following conditions Indicates the known coupler operation status The baseline characteristic mean, Represents the environmental sensitivity coefficient matrix. Represents the standard environment vector. Indicates the known coupler operation status and environmental condition vector Environmental characteristic covariance, Indicates the known coupler operation status The baseline characteristic covariance, This represents the adjustment coefficient matrix for environmental uncertainty. This represents the diagonalization operator.

[0112] Among them, the environmental sensitivity coefficient matrix This is used to characterize the degree of linear modulation of the mean of the trial response characteristics to changes in different environmental factors. This is because the internal friction, structural constraints, and the degree of involvement of elastic elements vary under different coupler operating conditions, thus resulting in differences in sensitivity to environmental factors such as temperature and humidity. Environmental uncertainty adjustment coefficient matrix. This is used to describe the impact of environmental changes on the discreteness and uncertainty expansion of the trial response characteristics. The diagonalization operator maps the changes in each environmental factor in the environmental condition vector to independent perturbation terms on the corresponding feature dimensions. and Statistical calibration was performed using offline experimental data collected under different operating conditions and environmental conditions.

[0113] pass and Construct a unified model for anticipating trial responses:

[0114] ;

[0115] In the formula, This indicates a trial response expectation model. Represents a multidimensional normal distribution. Indicates the known coupler operation status and environmental condition vector The mean environmental characteristics under the following conditions Indicates the known coupler operation status and environmental condition vector The environmental characteristics covariance.

[0116] Step S44: Perform feature-by-feature alignment comparison between the actual trial feature vector and the candidate trial response feature distribution, and calculate the consistency score of each trial response feature in the candidate trial response feature distribution;

[0117] In this step, because different test response features correspond to different physical mechanisms, their ability to distinguish coupler states and their environmental sensitivity vary. Directly using an overall distance metric could easily lead to feature masking or aberration. Therefore, a feature-by-feature consistency calculation method is adopted to achieve fine-grained alignment and independent evaluation at the feature level. Specifically, the conditional probability density value of the test response feature under the corresponding candidate test response feature distribution is calculated as the statistical consistency score for that feature.

[0118] ;

[0119] In the formula, Indicates the first Consistency score of each trial response feature Indicates the first A tentative response characteristic, and They represent the first The mean and covariance of the candidate tentative response feature distribution for each tentative response feature. This represents an exponential function.

[0120] Step S45: Perform multi-dimensional fusion of the consistency scores of all trial response features to obtain the matching credibility of the corresponding coupler candidate operation status.

[0121] In this step, the matching confidence level is used to characterize the degree to which the actual trial response, as a whole, conforms to the prior expectation of a certain coupler operation state under the current environmental conditions. The larger the value, the higher the statistical consistency between the candidate operation state and the current actual coupler state. Specifically, a weighted log-likelihood is used to perform multi-dimensional fusion of the consistency scores of all trial response features to obtain the matching confidence level of the corresponding coupler candidate operation state.

[0122] Step S5: Switch the operation mode of the unhooking robot based on the matching confidence level, and perform the unhooking operation based on the determined operation mode.

[0123] In this step, the candidate coupler operation state with the highest matching confidence is selected as the current coupler operation state. Based on the determined current coupler operation state, the operation mode of the uncoupling robot corresponding to the previous coupler operation state is selected from a predefined operation mode mapping table. Through the selected operation mode, the motion trajectory, speed planning, force control parameters, and safety constraints of the robot's end effector are adaptively adjusted to generate the corresponding uncoupling operation control command.

[0124] The unhooking operation is executed based on the generated unhooking operation control command, and the end position, contact force and motion status are monitored in real time during the execution process; when abnormal force response or operation failure characteristics are detected, the operation is interrupted or returned to the trial phase.

[0125] Example 2:

[0126] like Figure 2 As shown, this embodiment provides a multi-vehicle unhooking robot operation mode switching device, the device comprising:

[0127] The perception module is used to perceive the state of the target vehicle and obtain the vehicle type and coupler state vector. The coupler state vector includes the appearance features, spatial pose information and visible structure information of the coupler.

[0128] The generation module is used to generate multiple candidate coupler operation states based on the target vehicle model and the coupler state vector, combined with the structural prior of the current operation scenario.

[0129] The probing module is used to perform probing actions on the hook of the target vehicle through the unhooking robot and obtain probing interaction feedback data during the probing process;

[0130] The matching module is used to perform matching evaluation on each of the candidate coupler operation states based on the trial interaction feedback data, and obtain the matching credibility.

[0131] The switching module is used to switch the operation mode of the unhooking robot based on the matching confidence level, and to perform the unhooking operation based on the determined operation mode.

[0132] The sensing module includes:

[0133] The data acquisition unit is used to acquire the location information of the target vehicle within the work area and determine the uncoupling end of the target vehicle.

[0134] The first determining unit is used to obtain the vehicle model of the target vehicle based on the explicit identification information of the target vehicle or the prior information of the operating system.

[0135] The acquisition unit is used to acquire the prior spatial position of the coupler based on the type of the target vehicle.

[0136] The sensing unit is used to sense the state of the coupler of the target vehicle based on the coupler at the uncoupling operation end, given the prior spatial position of the coupler, and to obtain the coupler state vector.

[0137] The generation module includes:

[0138] The construction unit is used to construct the coupler operation state space based on the target vehicle model and the current operation scenario. The coupler operation state space includes the coupler models allowed for the vehicle model, coupler opening degree parameters, and coupler restricted state indicators.

[0139] The first screening unit is used to screen coupler configuration categories that are consistent with the appearance features in the coupler operation state space to obtain a candidate set of coupler configurations.

[0140] The derivation unit is used to perform geometric consistency constraints based on the spatial pose information of the key structure of the coupler, and to derive the feasible range of the coupler opening degree parameters.

[0141] The second determining unit is used to identify whether the coupler has structural features of restricted or abnormal release based on visible structural information, and to determine the candidate set of status indicators.

[0142] The second filtering unit is used to jointly constrain and filter the coupler operation state space based on the configuration candidate set, the feasible range of the coupler opening degree parameter, and the state identifier candidate set, so as to obtain multiple coupler candidate operation states that meet the current sensing conditions.

[0143] The probing module includes:

[0144] An initialization unit is used to initialize the trial action parameters based on the safety constraints of the current work scenario. The trial action parameters include the trial motion direction, the upper limit of the trial displacement, and the upper limit of the trial force.

[0145] The probing unit is used to control the unhooking robot to perform a probing action on the hook of the target vehicle along the probing motion direction based on the probing action parameters. The probing action includes a micro-displacement propulsion with a limited stroke and a controlled contact process.

[0146] The acquisition unit is used to acquire trial interaction feedback data through the contact component at the end of the unhooking robot during the execution of the trial action. The trial interaction feedback data includes displacement response data of the end displacement of the unhooking robot changing with time, data on the change of force at the end of the unhooking robot, and data on the change of damping during the contact process between the coupler and the contact component.

[0147] It should be noted that the specific manner in which each module performs its operation in the apparatus described in the above embodiments has been described in detail in the embodiments of the method, and will not be elaborated here.

[0148] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0149] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for switching operating modes of a multi-vehicle unhooking robot, characterized in that, include: The target vehicle is state-aware to obtain the vehicle type and coupler state vector. The coupler state vector includes the appearance features, spatial pose information and visible structure information of the coupler. Based on the vehicle type and the coupler state vector of the target vehicle, and combined with the structural prior of the current operation scenario, multiple candidate coupler operation states are generated. The unhooking robot performs a probing action on the hook of the target vehicle to obtain the probing interaction feedback data during the probing process; Based on the trial interaction feedback data, the matching and evaluation of each coupler candidate operation status is performed to obtain the matching credibility. The operation mode of the unhooking robot is switched based on the matching confidence level, and the unhooking operation is performed based on the determined operation mode; The step of using a hook-unhooking robot to perform a probing action on the hook of the target vehicle and obtaining probing interaction feedback data during the probing process includes: Based on the safety constraints of the current work scenario, initialize the trial action parameters, which include the trial motion direction, the upper limit of the trial displacement, and the upper limit of the trial force. The unhooking robot is controlled to perform a probing action on the hook of the target vehicle based on the probing action parameters along the probing motion direction. The probing action includes a micro-displacement propulsion with a limited stroke and a controlled contact process. During the execution of the probing action, probing interaction feedback data is collected through the contact component at the end of the unhooking robot. The probing interaction feedback data includes displacement response data of the end displacement of the unhooking robot changing with time, data on the change of force at the end of the unhooking robot, and data on the change of damping during the contact process between the coupler and the contact component. The step of matching and evaluating the candidate coupler operation status based on the trial interaction feedback data to obtain the matching reliability includes: By extracting features from the trial interaction feedback data, an actual trial feature vector including multiple trial response features is constructed; Obtain environmental information for the current work scenario; For each candidate coupler operation state, the candidate trial response feature distribution is obtained by combining the environmental information and using the trial response expectation model. The actual probe feature vector is compared with the candidate probe response feature distribution by feature alignment, and the consistency score of each probe response feature in the candidate probe response feature distribution is calculated. The consistency scores of all trial response features are fused in multiple dimensions to obtain the matching credibility of the corresponding coupler candidate operation status.

2. The method for switching operating modes of a multi-vehicle unhooking robot according to claim 1, characterized in that, The process of sensing the state of the target vehicle to obtain the vehicle type and coupler state vector includes: Collect the location information of the target vehicle within the work area and determine the uncoupling end of the target vehicle; Based on the explicit identification information of the target vehicle or the prior information of the operating system, obtain the model of the target vehicle; Based on the vehicle model, obtain the prior spatial position of the corresponding coupler; Given the prior spatial position of the coupler, the coupler state vector of the target vehicle is obtained by sensing the state of the coupler at the uncoupling end.

3. The method for switching operating modes of a multi-vehicle unhooking robot according to claim 2, characterized in that, Under the prior knowledge of the coupler's spatial position, the state perception of the target vehicle's coupler at the uncoupling operation end is performed to obtain the coupler state vector, including: At the end of the uncoupling operation, based on the prior spatial position of the coupler, a coupler sensing area is constructed in the working coordinate system of the uncoupling robot; Within the coupler sensing area, the appearance contour of the coupler is sampled to extract its appearance features; Based on the appearance features, the key structure of the coupler is extracted and spatially located to obtain spatial pose information. The spatial pose information includes the position coordinates of the key points of the coupler in the working coordinate system, the main axis direction and normal direction of the coupler, and the relative distance and relative height of the coupler relative to the uncoupling operation end. Based on the key structure and spatial pose information of the coupler, and combined with the prior knowledge of the coupler composition, the visible structural information of the coupler in the current state is identified.

4. The method for switching operating modes of a multi-vehicle unhooking robot according to claim 1, characterized in that, The generation of multiple candidate coupler operation states includes: Based on the vehicle type and current operating scenario, a coupler operation state space is constructed, which includes the coupler types allowed for the vehicle type, coupler opening degree parameters, and coupler restricted status indicators. In the coupler operation state space, coupler configuration categories that match the appearance features are filtered to obtain a candidate set of coupler configurations; Based on the spatial pose information of the key structure of the coupler, geometric consistency constraints are applied to derive the feasible range of the coupler opening degree parameters. Based on the visible structural information, identify whether the coupler has structural features that are restricted or abnormally released, and determine the candidate set of status indicators; Based on the candidate set of configurations, the feasible range of coupler opening degree parameters, and the candidate set of state identifiers, the coupler operation state space is jointly constrained and filtered to obtain multiple coupler candidate operation states that meet the current sensing conditions.

5. A multi-vehicle unhooking robot operation mode switching device, characterized in that, include: The perception module is used to perceive the state of the target vehicle and obtain the vehicle type and coupler state vector. The coupler state vector includes the appearance features, spatial pose information and visible structure information of the coupler. The generation module is used to generate multiple candidate coupler operation states based on the target vehicle model and the coupler state vector, combined with the structural prior of the current operation scenario. The probing module is used to perform probing actions on the hook of the target vehicle through the unhooking robot and obtain probing interaction feedback data during the probing process; The matching module is used to perform matching evaluation on each of the candidate coupler operation states based on the trial interaction feedback data, and obtain the matching credibility. The switching module is used to switch the operation mode of the unhooking robot based on the matching confidence level, and to perform the unhooking operation based on the determined operation mode; The probing module includes: An initialization unit is used to initialize the trial action parameters based on the safety constraints of the current work scenario. The trial action parameters include the trial motion direction, the upper limit of the trial displacement, and the upper limit of the trial force. The probing unit is used to control the unhooking robot to perform a probing action on the hook of the target vehicle along the probing motion direction based on the probing action parameters. The probing action includes a micro-displacement propulsion with a limited stroke and a controlled contact process. The acquisition unit is used to acquire trial interaction feedback data through the contact component at the end of the unhooking robot during the execution of the trial action. The trial interaction feedback data includes displacement response data of the end displacement of the unhooking robot changing with time, data of the change of force at the end of the unhooking robot, and data of damping change during the contact process between the coupler and the contact component. The step of matching and evaluating the candidate coupler operation status based on the trial interaction feedback data to obtain the matching reliability includes: By extracting features from the trial interaction feedback data, an actual trial feature vector including multiple trial response features is constructed; Obtain environmental information for the current work scenario; For each candidate coupler operation state, the candidate trial response feature distribution is obtained by combining the environmental information and using the trial response expectation model. The actual probe feature vector is compared with the candidate probe response feature distribution by feature alignment, and the consistency score of each probe response feature in the candidate probe response feature distribution is calculated. The consistency scores of all trial response features are fused in multiple dimensions to obtain the matching credibility of the corresponding coupler candidate operation status.

6. The multi-vehicle unhooking robot operation mode switching device according to claim 5, characterized in that, The sensing module includes: The data acquisition unit is used to acquire the location information of the target vehicle within the work area and determine the uncoupling end of the target vehicle. The first determining unit is used to obtain the vehicle model of the target vehicle based on the explicit identification information of the target vehicle or the prior information of the operating system. The acquisition unit is used to acquire the prior spatial position of the coupler based on the type of the target vehicle. The sensing unit is used to sense the state of the coupler of the target vehicle based on the coupler at the uncoupling operation end, given the prior spatial position of the coupler, and to obtain the coupler state vector.

7. The multi-vehicle unhooking robot operation mode switching device according to claim 5, characterized in that, The generation module includes: The construction unit is used to construct the coupler operation state space based on the target vehicle model and the current operation scenario. The coupler operation state space includes the coupler models allowed for the vehicle model, coupler opening degree parameters, and coupler restricted state indicators. The first screening unit is used to screen coupler configuration categories that are consistent with the appearance features in the coupler operation state space to obtain a candidate set of coupler configurations. The derivation unit is used to perform geometric consistency constraints based on the spatial pose information of the key structure of the coupler, and to derive the feasible range of the coupler opening degree parameters. The second determining unit is used to identify whether the coupler has structural features of restricted or abnormal release based on visible structural information, and to determine the candidate set of status indicators. The second filtering unit is used to jointly constrain and filter the coupler operation state space based on the configuration candidate set, the feasible range of the coupler opening degree parameter, and the state identifier candidate set, so as to obtain multiple coupler candidate operation states that meet the current sensing conditions.

Images