Chicken house unmanned carrying platform navigation positioning method based on under-cage feature extraction

Abstract

The invention relates to the technical field of navigation and positioning, in particular to a chicken coop unmanned carrying platform navigation and positioning method based on under-cage feature extraction, which comprises the following steps: scanning the edge of a trough under a cage through a linear array camera with an adjustable pitch angle, and extracting continuous geometric features based on Fourier descriptors; generating a characteristic datum line consistent with the extension direction of the cage body; performing kinematics binding on real-time motion parameters of the carrying platform and the characteristic datum line, and dynamically generating a navigation path curvature constraint according to the curvature change of the characteristic datum line; on the basis of the feature matching degree of the seam of the feeding trough between the adjacent cage bodies, the position and posture offset of the platform is calculated, the motion control parameters are reversely corrected, and closed-loop position and posture calibration is achieved. According to the method, the dynamic matching between the path curvature and the platform steering capability is realized, the physical feasibility and stability of path tracking are improved, and the operation requirements of narrow channels and large-curvature turning in a complex chicken house scene are met.

Description

Technical Field

[0001] The present invention relates to the technical field of navigation and positioning, and in particular to a navigation and positioning method for a chicken coop unmanned transport platform based on under-cage feature extraction. Background Art

[0002] With the intensive and automated development of the livestock and poultry breeding industry, links such as feeding, inspection, and manure cleaning inside the chicken coop increasingly rely on the participation of unmanned transport platforms. However, the internal structure of the chicken coop is generally dense and highly repetitive, with narrow space and complex lighting, and traditional navigation methods face many technical challenges: On the one hand, navigation methods based on global positioning systems (such as GPS, UWB) are vulnerable to interference from factors such as metal shielding and space closure in the chicken coop environment, resulting in signal distortion or failure, and it is difficult to meet the requirements of high-precision continuous navigation. On the other hand, navigation methods relying on ground path markings, two-dimensional codes, magnetic strips, or artificial anchor points have problems such as easy wear, high maintenance costs, and difficulty in adapting to dynamic environmental changes, and do not have the feasibility for long-term deployment.

[0003] In addition, although some current visual navigation methods can perform positioning by relying on environmental features, in the chicken coop, due to ground pollution, uneven lighting, and monotonous structure, the extraction of feature points is unstable, the positioning accuracy fluctuates greatly, and especially during the long-term operation of the platform, the cumulative error of the trajectory easily leads to pose drift, lacking an effective closed-loop correction mechanism. Summary of the Invention

[0004] The present invention provides a navigation and positioning method for a chicken coop unmanned transport platform based on under-cage feature extraction, which can adapt to the structure layout of the chicken coop, has a stable feature extraction ability and a high-robust path constraint control ability, and supports a navigation and positioning method for self-pose correction in a continuous operation state, so as to improve the operation efficiency and operation stability of the platform in a complex breeding environment.

[0005] A navigation and positioning method for a chicken coop unmanned transport platform based on under-cage feature extraction includes the following steps: S1: Scanning the edge of the under-cage feeder through a linear array camera with an adjustable pitch angle, extracting continuous geometric features based on Fourier descriptors, and generating a feature reference line consistent with the extension direction of the cage body; S2: Kinematically binding the real-time motion parameters of the transport platform to the feature reference line, and dynamically generating a navigation path curvature constraint according to the curvature change of the feature reference line; S3: Calculating the platform pose offset based on the feature matching degree at the feeder joint between adjacent cages, and reversely correcting the motion control parameters to achieve closed-loop pose calibration.

[0006] Optionally, the S1 specifically includes: S11. Adjust the pitch angle of the line array camera according to the height of the cage body and the installation inclination angle of the feeding trough, so that the scanning plane is orthogonal to the extension direction of the edge of the feeding trough; S12. Extract the set of pixel points on the edge of the feeding trough from the contour sequence images, and use Fourier descriptors to perform frequency-domain encoding on the edge contour. Retain the first N low-frequency coefficients to reconstruct the geometric curve, specifically including the set of pixel points on the edge extracted from each frame of the contour sequence images Perform frequency-domain encoding, using the discrete Fourier transform in complex form: , represents the horizontal coordinate of the th edge pixel point, represents the vertical coordinate of the th edge pixel point, is the imaginary unit, is the number of edge points, is the th Fourier coefficient. When reconstructing the geometric curve, retain the first frequency orders for reconstructing a smooth edge curve; S13. Based on the mechanical installation constraint relationship between adjacent cage bodies, perform spatial consistency verification on the reconstructed geometric curve, and screen out the continuous sections that meet the following conditions as the feature reference line: The distance error between the reference lines of adjacent cage bodies satisfies: ; The curvature change rate is less than the set curvature threshold: ; where, is the allowable tolerance for cage body installation, is the vertical distance deviation between adjacent reference lines, is the local curvature of the curve, is the curve arc length coordinate parameter, is the curvature change rate;

[0007] Optionally, the pitch angle of the line array camera is calculated as: ; where, is the height of the cage body, is the horizontal projection length of the feeding trough, is the installation inclination angle of the feeding trough, is the pitch angle that the line array camera needs to adjust.

[0008] Optionally, S1 further includes continuously collecting contour sequence images of the edge of the feeding trough by the line array camera along the traveling direction of the transport platform, and performing morphological closing operations on each frame of the contour sequence images to eliminate the interference of feed residues.

[0009] Optionally, S2 specifically includes: S21, obtaining the platform linear velocity in real time , the actual steering angle and the heading angle , establishing an Ackermann steering kinematic model of the carrier platform including the linear velocity, the actual steering angle and the heading angle, and obtaining the motion parameters in real time through the carrier platform, where the Ackermann steering kinematic model describes the geometric mapping relationship between the platform steering behavior and the front wheel angle; S22, extracting the obtained feature reference line, performing local geometric fitting on the feature reference line at the current position of the carrier platform, obtaining the corresponding desired path curvature, taking the real-time curvature of the feature reference line as the input of the desired path curvature, and converting the desired path curvature into the desired steering angle according to the kinematic model ; S23, according to the difference between the actual steering angle , adjusting the navigation path curvature constraint range ; S24, generating an allowable path curvature corridor based on the constraint range , and correcting the corridor boundary in combination with the platform maximum centripetal acceleration limit.

[0010] Optionally, S23 specifically includes: Based on , constructing the current acceptable navigation path curvature constraint range of the carrier platform, where the upper and lower boundaries of the constraint range are adjusted according to the steering response coefficient to reflect the actual response ability of the carrier platform under different speeds and adhesion conditions.

[0011] Optionally, in S22, the desired steering angle is calculated as: ; where is the desired steering angle, is the platform wheelbase, is the current curvature of the feature reference line, and the derivation is: .

[0012] Optionally, S3 specifically includes: S31, when the carrier platform travels to the adjacent cage joint, extracting the edge feature point set of the joint area and calculating its Hausdorff distance matching degree with the preset standard joint template ; S32, according to the deviation relationship between the matching degree and the reference calibration threshold, resolving the platform pose offset, including the lateral offset and the heading angle offset ; S33. Input and into the PID controller to generate the steering angular velocity compensation coefficient and the heading constraint correction amount ; S34. Reverse-correct the actual steering angle control parameter and the heading angle constraint parameter , and complete the closed-loop pose calibration.

[0013] Optionally, in S32, the solution of the platform pose offset is based on the deviation between the matching degree and the reference calibration threshold , expressed as: ; ; wherein, is the platform lateral position offset, is the platform heading angle offset, is the image error-physical displacement conversion coefficient, is the heading deviation correction coefficient, is the longitudinal physical span of the seam area feature, is the determination threshold of seam matching.

[0014] Advantages of the present invention: In the present invention, a linear array camera with an adjustable pitch angle scans the edge of the trough under the cage, and the edge curve is reconstructed in the frequency domain by combining Fourier descriptors, while eliminating feed interference and retaining a stable low-frequency contour shape, and a smooth section with a curvature change rate less than the threshold is selected as the feature reference line. Compared with the traditional navigation method based on global structure or single-point marking, this method can continuously and stably extract geometric navigation references in a closed environment with significant visual interference and high structural repetition, improving the continuity and anti-interference ability of platform positioning.

[0015] In the present invention, based on the real-time curvature of the feature reference line, a kinematic binding relationship of the Ackermann steering platform is established. An adaptive navigation curvature corridor is generated by the difference between the theoretical steering angle and the actual steering response, and physical boundary correction is performed by integrating the maximum centripetal acceleration constraint of the platform, effectively avoiding the problem of disconnection between path planning and platform dynamics, realizing the dynamic matching between path curvature and platform steering ability, improving the physical feasibility and stability of path tracking, and meeting the operation requirements of narrow channels and large-curvature turns in complex chicken coop scenarios.

[0016] In the operation process of the platform of the present invention, the feeding trough seam between adjacent cages is used as a natural positioning reference, the Hausdorff distance is introduced to perform template matching on the seam edge, and the lateral and course offsets of the platform are calculated based on the error magnitude, thereby forming an offset calculation mechanism based on the conversion between image and physical quantity. Combining with the compensation amount generated by the PID controller, the original navigation control parameters are corrected in reverse, realizing pose closed-loop self-calibration, overcoming the trajectory offset problem caused by visual drift or error accumulation in traditional navigation, and improving the long-term navigation accuracy and robustness in a large-scale chicken coop environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings described below are only those of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 It is a schematic flowchart of the method according to an embodiment of the present invention; Figure 2 It is a schematic diagram for extracting the feature reference line according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] The present invention will be described in detail below in conjunction with the drawings and specific embodiments. At the same time, it should be noted here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments. For some well-known technologies, those skilled in the art can also adopt other alternative methods for implementation; moreover, the drawings are only for more specifically describing the embodiments, and are not intended to specifically limit the present invention.

[0020] It should be pointed out that in the specification, it is mentioned that "an embodiment", "embodiments", "exemplary embodiments", "some embodiments", etc. indicate that the described embodiments may include specific features, structures or characteristics, but not necessarily every embodiment includes the specific feature, structure or characteristic. In addition, when combining an embodiment to describe a specific feature, structure or characteristic, implementing such a feature, structure or characteristic in combination with other embodiments (whether explicitly described or not) should be within the knowledge scope of those skilled in the relevant art.

[0021] Generally, terms can be understood, at least in part, from their use in context. For example, depending at least in part on the context, the term "one or more" as used herein can be used to describe any feature, structure, or property in a singular sense, or can be used to describe a combination of features, structures, or properties in a plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey a set of exclusive factors, but rather can alternatively, at least in part depending on the context, allow for the existence of other factors that are not necessarily explicitly described.

[0022] As Figure 1 - Figure 2 shown, a navigation and positioning method for an unmanned transport platform in a chicken coop based on under-cage feature extraction includes the following steps: S1: Scan the edge of the under-cage feeder with a linear array camera with adjustable pitch angle, extract continuous geometric features based on Fourier descriptors, and generate a feature reference line consistent with the extension direction of the cage body; S2: Kinematically bind the real-time motion parameters of the transport platform to the feature reference line, and dynamically generate a navigation path curvature constraint according to the curvature change of the feature reference line; S3: Calculate the platform pose offset based on the feature matching degree at the feeder joint between adjacent cages, and reverse-correct the motion control parameters to achieve closed-loop pose calibration.

[0023] S1 specifically includes: S11, pitch angle dynamic adjustment mechanism: According to the cage height and the installation inclination angle of the feeder , dynamically adjust the pitch angle of the linear array camera , so that the scanning plane is orthogonal to the extension direction of the feeder edge. The pitch angle calculation formula is: ; where is the cage height, is the horizontal projection length of the feeder, is the installation inclination angle of the feeder, is the pitch angle that the linear array camera needs to adjust.

[0024] S12, contour image anti-interference processing: The linear array camera continuously acquires a sequence of contour images of the feeder edge along the traveling direction of the transport platform, and performs morphological closing operations on each frame of the image to eliminate the interference of feed residues. The operation parameters include: The size of the closing operation structure element , where , that is, the maximum particle size of the feed pellets (measured value); The scanning synchronization relationship satisfies: , where is the line frequency of the linear array camera, is the running speed of the transport platform, is the physical space resolution corresponding to the pixel.

[0025] S13, Fourier descriptor feature extraction and reconstruction: The set of edge pixel points extracted from each frame of image (where ) is frequency-domain encoded, and the discrete Fourier transform in complex form is used: , represents the horizontal coordinate of the -th edge pixel point, represents the vertical coordinate of the -th edge pixel point, is the imaginary unit, which is used to represent the edge pixel points of the trough in complex form for the discrete Fourier transform in complex form; When reconstructing the curve, the first order low-frequency coefficients are retained, and the empirical value is: (corresponding reconstruction error < 0.5 mm), where, is the number of edge points, is the -th resampled complex-form edge point, is the -th Fourier coefficient, is the retained frequency order for reconstructing a smooth edge curve.

[0026] S14, Spatial consistency check and baseline screening: Based on the mechanical installation constraints of the cage body, the reconstructed geometric curve is subjected to a consistency check, and continuous sections that meet the following conditions are selected as the feature baseline: The spacing error between adjacent cage baselines satisfies: ; The curvature change rate is less than the set curvature threshold: ; where, is the allowable tolerance for cage installation, is the vertical distance deviation between adjacent baselines, is the local curvature of the curve, is the curve arc length coordinate parameter, is the curvature change rate (used to reflect the smoothness of the curve), and this threshold corresponds to the minimum turning radius .

[0027] S2 specifically includes: S21, Kinematic model binding: Establish an Ackermann steering kinematic model for the carrier platform, and obtain the linear velocity of the platform in real time 、actual steering angle and the heading angle , where the data source: the actual steering angle is provided by the in-vehicle CAN bus.

[0028] The Ackermann steering kinematic model is based on the non-slip assumption and is applicable to front-wheel steering, rear-wheel drive or four-wheel drive platforms. Its structure is defined as follows: I. Taking the center of mass as the reference point, model the two-dimensional motion of the platform in the plane. Assume that the front wheels rotate with the same steering angle while the rear wheels are parallel to the longitudinal axis of the platform and there is no side slip of the platform. The state variables defining the vehicle's center of mass include: the planar position (x, y), the heading angle ψ, the linear velocity v, the actual steering angle and the wheelbase Lwheelbase.

[0029] II. State equation (kinematic differential equation): The motion of the platform can be described by the following differential equations: The change in the center of mass position is determined by the linear velocity and the heading angle; The change in the heading angle is determined by the steering angle; The angular velocity of the platform is jointly affected by the linear velocity and the steering angle.

[0030] These equations are used for attitude prediction or path inversion during discrete control or simulation processes.

[0031] III. Parameter acquisition: Linear velocity v: Measured in real time by an encoder (installed on the drive wheel axle) and obtained by calculating the distance the wheel turns in unit time.

[0032] Actual steering angle : Measured in real time by a steering angle sensor or an angle potentiometer and read through the CAN bus or other communication interfaces; Heading angle ψ: Obtained by integrating the gyroscope in the inertial navigation system (IMU), or a more stable heading estimate can also be obtained by fusing GPS, magnetometer and odometer; After establishing this kinematic model, the actual , , values obtained from the sensors at any time can be compared with the reference values generated in path planning to perform trajectory tracking control.

[0033] S22, Calculation of the desired steering angle: Taking the real-time curvature of the feature reference line as the input of the desired path curvature, and solving the desired steering angle of the platform under this path curvature according to the Ackermann model: ; where is the desired steering angle, is the wheelbase of the platform (m),​ The current curvature of the feature reference line is derived as follows: .

[0034] S23, generation of the dynamic curvature constraint range: Calculate the difference between the actual steering angle and the desired steering angle: ; Adjust the path curvature constraint range according to this difference: ; Among them, is the platform steering response coefficient, is the allowable range of the navigation path curvature, and the platform steering response coefficient can be expressed as: , where is the platform linear velocity, is the gravitational acceleration (9.81 m / s 2 ), is the friction coefficient between the road surface and the wheels, dynamically adjusting the path curvature bandwidth to adapt to the response delay of the steering mechanism and the vehicle stability requirements.

[0035] S24, generating the allowable curvature corridor by fusing physical constraints: within the generated curvature constraint range , introduce the platform centripetal acceleration limit and correct the boundary: , where is the maximum path curvature that the platform can withstand, ensuring that the platform will not slip or become unstable due to excessive centrifugal force when following the curvature change. The finally generated navigation path curvature corridor will be , which is used for the controller to dynamically generate a feasible navigation path.

[0036] The output of S1 is the geometric reference line segment that has been smoothed and reconstructed by the Fourier descriptor and screened through the consistency check, that is: a two-dimensional continuous curve segment representing the edge contour of the trough in the image space, with a small curvature fluctuation and a high-confidence navigation reference line representing the extension direction of the cage body; the input of S2 is the curvature value corresponding to each local window or the current frame in this reference line , as the desired path curvature.

[0037] The acquisition method of is as follows: The feature reference line is a curve composed of a series of smooth pixel points, and the local curvature can be calculated through these points: ; where represents the two-dimensional coordinate points on the reference line, parameterized by the arc length is the first derivative, representing the tangent direction, is the second derivative, representing the change in the curvature direction. These derivative values are obtained by numerically differentiating through polynomial fitting (cubic spline) of the reference line.

[0038] In S1 evaluates the curvature change of the entire curve at multiple consecutive points, and is used to screen for sections with stable curvature. In S2 extracts the curvature at the current position of the platform as the control input from the stable curve segments screened by S1. If is small, it indicates that the of this section of the curve is smooth and continuous during platform movement, and is suitable as a navigation path; at this time, the value extracted from any point in this section (such as the current position of the platform) is more representative, with smaller errors and more stable control. Therefore is the curvature value of the navigation input point selected under the condition of meeting the threshold condition.

[0039] S3 specifically includes: S31, calculation of feature seam matching degree: when the carrier platform moves to the adjacent cage seam area, extract the edge feature point set of this area , and compare it with the preset standard seam template feature point set and use the Hausdorff distance for matching degree measurement: ; Among them, is the Hausdorff distance matching degree (pixels) between the current seam area and the template, is the seam edge feature point set extracted in the current frame, is the standard template feature point set, is the Euclidean distance.

[0040] S32, calculation of platform pose offset: according to the deviation between the matching degree and the reference calibration threshold set by the system, calculate the lateral offset and the heading angle offset of the platform: ; ; Among them, is the lateral position offset of the platform, is the heading angle offset of the platform, is the image error - physical displacement conversion coefficient, is the course deviation correction coefficient. Under the conditions that the conventional speed of the carrier platform is 0.4–0.6 m / s and the attitude control response time < 0.5 s, the optimal correction result is , is the longitudinal physical span of the seam area feature, is the decision threshold for seam matching. is used to convert the Hausdorff distance (image matching deviation in pixels) into the actual lateral physical offset of the platform. The acquisition method is through a calibration experiment: arrange multiple physical samples of cage seams with known lateral offsets on both sides of the track ( ); use the linear array camera installed on the platform to collect the image seam features of each sample at a fixed attitude; calculate the Hausdorff distance for each group of samples ; for each known offset and the corresponding to establish a set of calibration data points; perform a unary linear regression: . The fitting slope is , and the residual is controlled within an acceptable mm. In the actual measurement of the present invention, the imaging resolution of the linear array camera is 0.25 mm / pixel. After calibration fitting, we get: .

[0041] The determination method of the reference calibration threshold is as follows: collect 100 groups of real seam images in a controlled environment, including: 80 groups of standard seam samples with intact structures and no offsets, and 20 groups of non-standard samples with known offsets or tilt errors. Calculate the Hausdorff distance between all samples and the standard template , and obtain the maximum distance of the standard group and the minimum distance of the non-standard group respectively. Take the median value of the two as the classification decision threshold: .

[0042] Experimental sample results (actually measured): maximum distance of standard samples: pixels; minimum distance of non-standard samples: pixels. Therefore, the matching decision threshold is: .

[0043] S33, generation of steering and course compensation parameters: for the offset and Input a PID controller to generate the steering angle compensation coefficient respectively and the heading angle correction amount : ; ; wherein, is the multiplicative compensation factor of the steering angular velocity, is the additive correction amount of the heading angle, are the proportional, integral and differential gain coefficients of the PID controller respectively, is the heading angle correction proportional coefficient, is the original heading angle.

[0044] S34, Closed-loop correction of navigation control parameters: Based on the above compensation amounts, correct the actual steering angle control parameter and the heading angle constraint parameter to obtain the control parameters after closed-loop correction: ; ; wherein, is the target steering angle after closed-loop correction, is the target heading angle after closed-loop correction, and the above two quantities will be fed back to the controller to achieve dynamic correction of the path deviation.

[0045] The specific construction of the preset standard seam template feature point set is as follows: 1. Acquisition method: The standard seam template feature point set is obtained by selecting images of multiple actual cage seam areas with complete structures and high installation accuracies under the condition after the mechanical installation of the cage body, and using Canny edge detection + morphological filtering to extract edge features.

[0046] 2. Point set structure: The standard template is a two-dimensional coordinate point set, denoted as: wherein, is the number of feature points in the template, and 50 - 100 representative edge points are selected. The feature points cover the typical geometric shapes of the seam line segment in the image, such as vertical broken seams, edge transition inflection points, etc.; all point sets are subjected to coordinate normalization and smoothing processing to improve the matching stability and noise resistance.

[0047] 3. Feature alignment strategy: Use centroid alignment and principal direction alignment (PCA direction) methods to perform pre-registration processing on the template and the current image seam area to ensure that the Hausdorff distance reflects the true geometric deviation.

[0048] The present invention covers any alternatives, modifications, equivalent methods, and solutions within the spirit and scope of the present invention. To enable the public to have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention. However, those skilled in the art can fully understand the present invention even without the description of these details. Additionally, well-known methods, processes, procedures, components, and circuits are not described in detail to avoid unnecessary confusion with the essence of the present invention.

[0049] The above description is only a preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, several improvements and refinements can be made without departing from the principle of the present invention, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A navigation and positioning method for an unmanned transport platform in a chicken coop based on under-cage feature extraction, characterized in that, Including the following steps: S1: Scanning the edge of the trough under the cage by a linear array camera with adjustable pitch angle, extracting continuous geometric features based on Fourier descriptors, and generating a feature reference line consistent with the extending direction of the cage body; S2: Kinematically binding the real-time motion parameters of the carrier platform with the feature reference line, and dynamically generating a curvature constraint for the navigation path according to the curvature change of the feature reference line; S3: Calculating the platform pose offset based on the feature matching degree at the trough joint between adjacent cages and inversely correcting the motion control parameters to achieve closed-loop pose calibration.

2. The navigation and positioning method of the chicken coop unmanned carrier platform based on under-cage feature extraction according to claim 1, wherein, The specific content of S1 includes: S11: Adjusting the pitch angle of the linear array camera according to the cage height and the installation inclination angle of the trough, so that the scanning plane is orthogonal to the extending direction of the trough edge; S12. Extract the set of pixel points on the edge of the feeding trough from the contour sequence images, perform frequency-domain encoding on the edge contour using Fourier descriptors, and retain the first N low-frequency coefficients to reconstruct the geometric curve, specifically including the set of pixel points on the edge extracted from each frame of the contour sequence images Perform frequency-domain encoding using the discrete Fourier transform in complex form: represents the horizontal coordinate of the th edge pixel point, represents the vertical coordinate of the th edge pixel point, is the imaginary unit, is the number of edge points, is the rd Fourier coefficient. When reconstructing the geometric curve, retain the first frequency orders for reconstructing a smooth edge curve; S13: Based on the mechanical installation constraint relationship between adjacent cages, performing a spatial consistency check on the reconstructed geometric curve, and screening out continuous sections that meet the following conditions as the feature reference line: The spacing error between the reference lines of adjacent cages satisfies: ; Rate of change of curvature Less than the set curvature threshold: ; Among them, is the allowable tolerance for cage installation, is the vertical distance deviation between adjacent reference lines, is the local curvature of the curve, is the curve arc length coordinate parameter, is the curvature change rate.

3. A navigation and positioning method for a chicken coop unmanned carrier platform based on under-cage feature extraction according to claim 2, characterized in that The pitch angle of the linear array camera is calculated as follows: ; where is the height of the cage body, is the horizontal projection length of the feeding trough, is the installation inclination angle of the feeding trough, is the pitch angle that the linear array camera needs to adjust.

4. A navigation and positioning method for a chicken coop unmanned transport platform based on under-cage feature extraction according to claim 2, characterized in that, S1 also includes continuously collecting contour sequence images of the trough edge by the linear array camera along the traveling direction of the carrier platform, and performing morphological closing operations on each frame of the contour sequence images to eliminate the interference of feed residues.

5. A navigation and positioning method for a chicken coop unmanned transport platform based on under-cage feature extraction according to claim 1, characterized in that The specific content of S2 includes: S21, Obtain the platform linear velocity in real time , the actual steering angle and the heading angle , establish an Ackermann steering kinematic model of the carrier platform including the linear velocity, the actual steering angle and the heading angle, and obtain the motion parameters in real time through the carrier platform, where the Ackermann steering kinematic model describes the geometric mapping relationship between the platform steering behavior and the front wheel angle; S22. Extract the obtained feature reference line, perform local geometric fitting on the feature reference line at the current position of the carrier platform, obtain the corresponding expected path curvature, and use the real-time curvature of the feature reference line as the input of the expected path curvature. According to the kinematic model, convert the expected path curvature into the expected steering angle ; S23. Adjust the navigation path curvature constraint range according to the difference between the actual steering angle and ; ;​ S24, based on the constraint range Generate an allowable path curvature corridor and correct the corridor boundary by combining the platform's maximum centripetal acceleration limit.

6. A navigation and positioning method for a chicken coop unmanned carrier platform based on under-cage feature extraction according to claim 5, characterized in that The S23 specifically includes: Based on , construct the curvature constraint range of the navigation path currently acceptable to the carrier platform, and the upper and lower boundaries of the constraint range are adjusted according to the steering response coefficient to reflect the actual response ability of the carrier platform under different speeds and adhesion conditions.

7. A navigation and positioning method for a chicken coop unmanned transport platform based on under-cage feature extraction according to claim 5, characterized in that In S22, the desired steering angle is calculated as follows: ; wherein, is the desired steering angle, is the platform wheelbase, is the current curvature of the characteristic reference line, and is derived as: 。 8. A navigation and positioning method for a chicken coop unmanned transport platform based on under-cage feature extraction according to claim 1, characterized in that The specific content of S3 includes: S31. When the carrying platform moves to the joint of adjacent cages, extract the set of edge feature points in the joint area and calculate the Hausdorff distance matching degree between it and the preset standard joint template. ; S32, according to the matching degree and the deviation relationship with the reference calibration threshold, calculate the platform pose offset, including the lateral offset and the heading angle offset ; S33, input and into the PID controller to generate the steering angular velocity compensation coefficient and the heading constraint correction amount ; S34, Reverse-correct the actual steering angle control parameter and the heading angle constraint parameter , and complete the closed-loop pose calibration.

9. A navigation and positioning method for a chicken coop unmanned carrier platform based on under-cage feature extraction according to claim 8, wherein, In S32, the calculation of the platform pose offset is based on the deviation of the matching degree from the reference calibration threshold and is expressed as: ； ； Among them, is the lateral position offset of the platform, is the course angle offset of the platform, is the image error - physical displacement conversion coefficient, is the course deviation correction coefficient, is the longitudinal physical span of the seam area feature, is the determination threshold for seam matching.

Images