Orchard transportation robot path tracking system

Abstract

The invention discloses an orchard transportation robot path tracking system, and relates to the technical field of robot programmes, and the system comprises the steps: collecting an alignment data set, extracting a stability feature, a terrain trafficability feature and initial trajectory deviation data based on the alignment data set, generating a deviation scoring result based on the initial trajectory deviation data and the trafficability feature, and carrying out the tracking of an orchard transportation robot path. According to the method, autonomous path tracking and avoidance control of the orchard transportation robot are realized through multi-source perception and intelligent decision, positioning, attitude and environment structure data are fused, the advancing stability and terrain trafficability are analyzed in real time, and the accuracy and the reliability of the orchard transportation robot are improved. The path deviation trend is predicted and adaptively corrected, high-precision navigation in a complex orchard environment is realized, the robot can keep stable operation in a multi-machine cooperation and edge-of-field steering scene through a dynamic channel regulation and control and feedback learning mechanism, and automation and operation efficiency of orchard transportation operation are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of robot programming, and in particular to a path tracking system for orchard transport robots. BACKGROUND

[0002] In recent years, with the continuous improvement of people's health awareness, sea buckthorn fruits rich in vitamin C have gradually become a new favorite for health preservation. However, the transportation link after sea buckthorn fruit picking faces challenges such as low efficiency, high cost, and difficulty in coping with complex field environments, which seriously restricts the improvement of its quality guarantee and economic benefits.

[0003] At present, the Chinese invention patent with application number CN202411610564.7 discloses a robot dynamic path tracking method and system, which comprises: for a differential robot chassis, a nonlinear model predictive control model is constructed; the obstacle environment information around the robot is considered to construct a target cost function; the control input vector in the nonlinear model predictive control model is adjusted to perform real-time tracking control on the robot, the target cost function value is calculated, and the control input vector corresponding to the minimum target cost function value is the optimal control input vector. The optimal control input vector is used for dynamic path tracking control of the robot.

[0004] The above-mentioned technology is difficult to realize real-time prediction, autonomous correction and multi-robot collaborative avoidance control of path deviation of the transport robot in the complex and unstructured environment of the orchard. SUMMARY

[0005] The technical problem solved by the present application is that the prior art is difficult to realize real-time prediction, autonomous correction and multi-robot collaborative avoidance control of path deviation of the transport robot in the complex and unstructured environment of the orchard.

[0006] To solve the above technical problems, the present application provides the following technical solutions: An orchard transport robot path tracking system, comprising an environment acquisition alignment module, a travel feature analysis module, a deviation prediction and scoring module, a tracking strategy generation module, a dynamic channel regulation module, and a feedback correction module; The environment acquisition alignment module is used to acquire the position information and environment structure information of the robot when running in the rows of the orchard, and to perform synchronous processing based on the master clock to generate an aligned data set; The travel feature analysis module is used to extract stability features, terrain passability features and initial trajectory deviation data based on the aligned data set; The deviation prediction and scoring module is used to generate a deviation score result based on the initial trajectory deviation data and the passability features; The tracking strategy generation module is used to generate candidate strategy parameters according to the deviation score result, select the optimal strategy and perform control; The dynamic channel regulation module is used for avoidance and path switching in the orchard multi-robot simultaneous operation and path end scene; The feedback correction module is used for collecting execution feedback data and updating the score threshold and parameter weight.

[0007] Preferably, the environment collection alignment module includes a position and attitude observation unit, a row and column structure recognition unit, and a time alignment unit; The position and attitude observation unit is used for collecting satellite positioning data, heading angle data, and attitude angle data, and performing same-frequency collection and validity test on the satellite positioning data, heading angle data, and attitude angle data, and outputs position and attitude observation data; The row and column structure recognition unit is used for obtaining external scene information, and recognizing the direction between fruit tree rows, the effective width of the channel, and the left and right boundary positions based on the arrangement characteristics of the fruit tree planting rows; The collected scene data are subjected to spatial feature extraction and geometric morphology analysis, and based on the tree crown density variation, row offset trend, and row bending characteristics, it is determined whether the row and column direction of the fruit trees has local narrowing, turning, or offset, and outputs row and column structure data; The time alignment unit is used for time synchronization processing of the position and attitude observation data and the row and column structure data based on the master clock, and outputs an aligned data set.

[0008] Preferably, the travel feature analysis module includes a stability feature extraction unit, a terrain passability analysis unit, and a baseline trajectory deviation calculation unit; The stability feature extraction unit is used for analyzing the running stability of the robot in the channel between the fruit tree rows based on the aligned data set, and through joint determination of the heading angle change curve, attitude angle fluctuation interval, and travel speed continuity, it identifies whether the robot has a tendency to yaw, a lateral offset trend, or a decrease in travel stability, and outputs change features, which are subjected to smoothing processing, period analysis, and trend extraction, and output stability feature data; The terrain passability analysis unit extracts surface feature variation and attitude feature from the aligned data set, and outputs passability factor data; The baseline trajectory deviation calculation unit calculates the lateral position difference and the heading direction difference based on the corresponding relationship between the satellite positioning data and the reference path center axis in the same coordinate system, and outputs initial trajectory deviation data; The initial trajectory deviation data is used to represent the numerical form output of the deviation of the current position of the robot relative to the ideal travel path; The reference path center axis is determined based on the channel center line obtained by the row and column structure recognition unit and the work path planning model.

[0009] Preferably, the deviation prediction scoring module comprises a trajectory deviation trend prediction unit, a passability correction unit and a deviation scoring unit; The trajectory deviation trend prediction unit is configured to analyze the trend of trajectory deviation over time based on initial trajectory deviation data and stability feature data, identify whether the deviation is in an aggravating, alleviating or stable state by jointly comparing the lateral offset, the heading angle offset and the attitude angle change rate, and output predicted deviation data; The trajectory deviation trend prediction unit performs smoothing processing on instantaneous amplitude mutations and trend fitting on continuous deviation paragraphs during the analysis process; The passability correction unit is configured to correct the predicted deviation data according to passability factor data, and the correction process is performed by associating the predicted deviation data with the passability factor data. When the passability factor shows that the terrain resistance is large or the instability increases, the sensitivity to the change of the predicted deviation is increased to a first sensitivity value, and corrected deviation data is output; The deviation scoring unit is configured to perform weighted synthesis on the corrected deviation data to generate a deviation score result.

[0010] Preferably, the tracking strategy generation module comprises a candidate strategy generation unit, an optimal strategy selection unit and an execution control unit; The candidate strategy generation unit is configured to divide the deviation score result into intervals based on the deviation score result, and generate a candidate strategy parameter set, the candidate strategy parameter set comprising a look-ahead distance adjustment amount and a speed adjustment amount, wherein the look-ahead distance adjustment amount is used to represent the distance of a target point in front of the path referred to by the robot during travel, and the speed adjustment amount is used to represent the change amplitude of the travel speed of the robot under the current passability condition; The optimal strategy selection unit compares the candidate strategy parameter set according to the deviation score result and the passability factor; Calculate the initial trajectory deviation difference before and after the change of the look-ahead distance adjustment amount, and take the initial trajectory deviation difference as deviation correction effect data; Calculate the stability feature data difference before and after the change of the speed adjustment amount, and take the stability feature data difference as through stability impact data; Compare the deviation correction effect data and the through stability impact data: When the deviation correction effect data is greater than the change amount of the through stability impact data, select the strategy parameter whose look-ahead distance adjustment amount is greater than a preset look-ahead distance threshold value; When the through stability impact data is greater than the change amount of the deviation correction effect data, select the strategy parameter whose speed adjustment amount is greater than a preset speed adjustment amount threshold value; output the selected strategy parameters as optimal strategy parameters; The execution control unit is configured to adjust the wheel speed difference and the forward direction according to the optimal strategy parameters, and output execution control data.

[0011] Preferably, the dynamic channel regulation module comprises a meeting identification unit, a meeting regulation unit, and a head-of-channel turning guidance unit. The meeting identification unit reads the position, direction of travel, and speed data of the current robot and other robots from the alignment data set, calculates the relative distance according to the position difference, calculates the relative speed according to the speed difference, and forms time-to-collision data through the ratio of the relative distance and the relative speed. The time-to-collision data is compared with channel geometric constraint data formed by the effective width of the channel and the crown obstruction to determine whether both can pass through at the same time under the current channel conditions, and meeting identification data is generated. The meeting regulation unit is configured to select a waiting, insertion, or yielding strategy according to the meeting identification data and the deviation score result. After judging the position relationship between the robot and other robots, the passable factor, and the comprehensive result of the path deviation score: If the meeting identification data indicates a potential collision risk and the deviation score is lower than a preset deviation score threshold, a waiting strategy is selected. If it is necessary to form a passing order in a multi-robot flow, an insertion strategy is selected. When the deviation score is higher than the preset deviation score threshold and the channel conditions allow, a yielding strategy is selected. The head-of-channel turning guidance unit is configured to perform a path switching action at the end of the path, and the path switching action is: A transition posture is generated in advance before approaching the end of the path, a pre-swing adjustment is made to an alignment turning initial posture, a turning action is completed by controlling the wheel speed difference according to the set curve curvature of the bend, and the robot is realigned to the center axis of the next reference path through a merging process.

[0012] Preferably, the feedback correction module comprises a feedback data acquisition unit, a threshold updating unit, and a weight adjustment unit. The feedback data acquisition unit is configured to acquire trajectory deviation, posture fluctuation, and passing stability data generated during the robot's operation after the path tracking strategy is executed, perform time series recording and effectiveness checking on the trajectory deviation, posture fluctuation, and passing stability data, and generate feedback data. The threshold updating unit analyzes the change trend of the trajectory deviation and the posture fluctuation in the feedback data, judges whether the current deviation score threshold can accurately reflect the actual operation state of the robot, and outputs a deviation score threshold adjustment signal when the feedback data indicates that there is a response deficiency. The weight adjustment unit compares and analyzes the influence of different characteristic changes in the feedback data on the operation effect, determines the characteristic influence that should be enhanced or weakened in the subsequent deviation prediction and scoring process, and outputs a correction parameter set.

[0013] Preferably, the deviation scoring unit further comprises a dynamic weight adjustment mechanism. The dynamic weight adjustment mechanism is used to adjust the weight proportion of the stability characteristic and the passability factor in the deviation scoring in real time according to the change trend of different orchard terrains and operation conditions.

[0014] Preferably, the meeting control unit preferentially selects the path integration strategy when it is judged that the meeting conflict is likely to continuously increase.

[0015] Preferably, the threshold updating unit further combines the execution control data and the time difference for hysteresis compensation when updating the deviation scoring threshold.

[0016] The present application has the following beneficial effects: The present application realizes autonomous path tracking and avoidance control of the orchard transport robot through multi-source perception and intelligent decision making, fuses positioning, attitude and environmental structure data, analyzes the travel stability and terrain passability in real time, predicts the path deviation trend and adaptively corrects it, realizes high-precision navigation in complex orchard environment, and through the dynamic channel regulation and feedback learning mechanism, the robot can maintain stable operation in the multi-machine cooperation and the ground turning scene, and significantly improves the automation and operation efficiency of the orchard transportation operation. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 A basic flowchart of an orchard transport robot path tracking system is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments.

[0019] Embodiments, with reference to Figure 1 , an orchard transport robot path tracking system is provided, which comprises an environment acquisition alignment module, a travel characteristic analysis module, a deviation prediction and scoring module, a tracking strategy generation module, a dynamic channel regulation module and a feedback correction module.

[0020] The environment acquisition alignment module is used to acquire the position information and environmental structure information of the robot when running in the orchard row, and synchronously processes based on the master clock to generate an alignment data set.

[0021] The traveling feature analysis module is used to extract stability features, terrain passability features, and initial trajectory deviation data based on the alignment dataset.

[0022] The deviation prediction scoring module is used to generate deviation score results based on the initial trajectory deviation data and the passability features.

[0023] The tracking strategy generation module is used to generate candidate strategy parameters according to the deviation score results, select the optimal strategy, and perform control.

[0024] The dynamic channel regulation module is used to avoid and switch paths in the orchard multi-robot simultaneous operation and path end scene.

[0025] The feedback correction module is used to collect execution feedback data and update the scoring threshold and parameter weight.

[0026] With the rapid development of new generation information technologies such as artificial intelligence and machine vision, the intelligent and automated level of agricultural production is continuously improving, and the performance requirements for agricultural robots are becoming higher and higher. At the same time, in the transportation link after sea buckthorn fruit picking, the complex and variable field environment poses great challenges to the adaptability and reliability of transportation equipment. The sea buckthorn field transportation robot developed in this project integrates machine vision, GPS navigation and autonomous obstacle avoidance technology, and can realize the automatic and intelligent transportation of sea buckthorn fruit.

[0027] The research hypothesis of the present invention is: (1) A fruit orchard transportation robot path tracking system based on machine vision, GPS navigation and autonomous obstacle avoidance technology can effectively solve the problems of low efficiency, high cost and poor adaptability of traditional manual transportation methods.

[0028] (2) The system can realize the automatic and intelligent transportation of sea buckthorn fruit, improve the transportation efficiency, reduce the labor cost, and effectively protect the quality of sea buckthorn fruit.

[0029] (3) The application of the system will promote the mechanization and intelligent development of the sea buckthorn fruit industry, improve the industrial competitiveness, promote the income of farmers, and ultimately realize the sustainable development of the sea buckthorn fruit industry.

[0030] The successful implementation of this project will provide an efficient, intelligent and environmentally friendly transportation solution for the sea buckthorn fruit industry, which is of great significance for promoting the high-quality development of the sea buckthorn fruit industry, promoting agricultural modernization and realizing rural revitalization.

[0031] The application realizes autonomous path tracking and avoidance control of the orchard transport robot through multi-source perception and intelligent decision-making, fuses positioning, attitude and environmental structure data, analyzes the travel stability and terrain passability in real time, predicts the path deviation trend and adapts to correction, realizes high-precision navigation in complex orchard environment, and through dynamic channel regulation and feedback learning mechanism, the robot can keep stable operation in multi-machine cooperation and ground turning scene, significantly improving the automation and operation efficiency of orchard transportation operation.

[0032] The environment acquisition alignment module comprises a position and attitude observation unit, a row and column structure identification unit and a time alignment unit.

[0033] The position and attitude observation unit is used for collecting satellite positioning data, heading angle data and attitude angle data, and performing same-frequency collection and validity test on the satellite positioning data, the heading angle data and the attitude angle data, and outputs position and attitude observation data.

[0034] The position and attitude observation unit realizes continuous collection and validity test of satellite positioning, heading and attitude, and ensures that the running position data of the robot in the orchard is accurate and stable.

[0035] The row and column structure identification unit is used for acquiring external scene information, and identifying the direction between fruit tree rows, the effective width of the channel and the left and right boundary positions based on the arrangement characteristics of the fruit tree planting rows.

[0036] The row and column structure identification unit is used for acquiring external scene information, and identifying the direction between fruit tree rows, the effective width of the channel and the left and right boundary positions based on the arrangement characteristics of the fruit tree planting rows.

[0037] The row and column structure identification unit identifies the channel boundary and direction through the arrangement of fruit trees and visual geometric characteristics, and generates accurate channel structure information.

[0038] The time alignment unit is used for time synchronization processing of the position and attitude observation data and the row and column structure data based on the master clock, and outputs an aligned data set.

[0039] The time alignment unit performs time synchronization and smooth interpolation on the multi-source data based on the master clock, and ensures that the position and environmental information completely correspond in time sequence.

[0040] Through the environment acquisition alignment module, the system can accurately acquire the spatial position, attitude state and channel geometric characteristics of the robot in the complex scene between the fruit tree rows, and synchronously integrate the multi-source data into a unified aligned data set, realize the integration of positioning, identification and time registration, and provide high-reliability input for path deviation analysis and strategy decision.

[0041] The travel feature analysis module comprises a stability feature extraction unit, a terrain passability analysis unit and a baseline trajectory deviation calculation unit.

[0042] The stability feature extraction unit is configured to analyze the running stability of the robot in the inter-row channel of the fruit trees based on the aligned data set, identify whether the robot has a tendency of yawing, a trend of lateral deviation or a decrease in travel stability by jointly determining the change curve of the heading angle, the fluctuation range of the attitude angle and the continuity of the travel speed, and output change features. The change features are smoothed, analyzed for periodicity and extracted for trends, and stability feature data is output.

[0043] The stability feature extraction unit identifies the yawing, lateral deviation or attitude fluctuation trend of the robot during running by analyzing the dynamic changes of the heading angle, the attitude angle and the travel speed, and outputs stability feature data that can reflect the running stability, thereby improving the real-time perception ability of the system for the travel state.

[0044] The terrain passability analysis unit extracts the surface feature change and the attitude feature from the aligned data set, and outputs passability factor data.

[0045] The terrain passability analysis unit extracts the resistance and passability index of the terrain to the robot travel by analyzing the surface fluctuation and the attitude change, and forms the passability factor data to provide terrain constraint information for the path deviation correction.

[0046] The baseline trajectory deviation calculation unit calculates the lateral position difference and the heading direction difference based on the corresponding relationship between the satellite positioning data and the reference path center axis in the same coordinate system, and outputs initial trajectory deviation data.

[0047] The initial trajectory deviation data is used to represent the deviation degree of the current position of the robot relative to the ideal travel path in the form of numerical output.

[0048] The reference path center axis is determined based on the channel center line obtained by the row and column structure recognition unit and the work path planning model.

[0049] The baseline trajectory deviation calculation unit maps the current position of the robot and the reference path center axis in space, calculates the lateral and heading differences, and outputs the initial trajectory deviation data, thereby realizing quantitative representation of the path deviation degree and laying an accurate foundation for subsequent deviation prediction and control strategies.

[0050] The travel feature analysis module realizes quantitative analysis of the running attitude of the robot, the terrain adaptability and the path deviation by comprehensive calculation on the aligned data set. The module converts the multi-source sensing data into a feature set that can be used for trajectory prediction and path control, so that the system can judge the stability degree of the running state of the robot, the terrain passability and the deviation from the ideal path in real time, and provide accurate basis for subsequent strategy generation.

[0051] The deviation prediction scoring module comprises a trajectory deviation trend prediction unit, a passability correction unit, and a deviation scoring unit.

[0052] The trajectory deviation trend prediction unit is configured to analyze the trend of trajectory deviation over time based on initial trajectory deviation data and stability feature data, identify whether the deviation is in an aggravating, alleviating, or stable state by jointly comparing the lateral offset, the heading angle offset, and the attitude angle change rate, and output predicted deviation data.

[0053] The trajectory deviation trend prediction unit performs smoothing processing on instantaneous amplitude mutations and trend fitting on continuous deviation paragraphs during the analysis process.

[0054] The trajectory deviation trend prediction unit identifies the development trend of path deviation by jointly analyzing the lateral offset, the heading angle offset, and the attitude angle change rate, performs smoothing and trend fitting on mutation data, and outputs predicted deviation data, so that the system can predict the direction and rate of deviation change.

[0055] The passability correction unit is configured to correct the predicted deviation data according to passability factor data. The correction process is performed by associating the predicted deviation data with the passability factor data. When the passability factor indicates that the terrain resistance is large or the instability increases, the sensitivity to the predicted deviation change is increased to a first sensitivity value, and corrected deviation data is output.

[0056] The passability correction unit adjusts the sensitivity of the predicted deviation data based on the terrain passability factor. When a terrain resistance or instability region is detected, the deviation response sensitivity is increased, so that the prediction result is closer to the actual passability state.

[0057] The deviation scoring unit is configured to perform weighted synthesis based on the corrected deviation data to generate a deviation score result.

[0058] The deviation scoring unit further comprises a dynamic weight adjustment mechanism.

[0059] The dynamic weight adjustment mechanism is configured to adjust the weight proportion of the stability feature and the passability factor in the deviation scoring in real time according to the change trend of different orchard terrains and operating conditions.

[0060] The deviation scoring unit performs weighted calculation on the corrected deviation data, considers the deviation amplitude, change trend, and terrain influence, and outputs a deviation score result, which provides quantitative input for subsequent path tracking strategy generation.

[0061] The deviation prediction scoring module realizes quantitative evaluation and dynamic prediction of the robot path deviation by analyzing and correcting the trajectory deviation change trend. The module comprehensively considers the attitude change, terrain resistance and traffic stability, so that the system can identify the path deviation trend in advance and generate a quantifiable deviation score result, providing a forward-looking decision basis for path control and improving the path keeping accuracy and running stability of the orchard robot in complex environments.

[0062] The tracking strategy generation module includes a candidate strategy generation unit, an optimal strategy selection unit and an execution control unit.

[0063] The candidate strategy generation unit is configured to divide the deviation score result into intervals based on the deviation score result, and generate a candidate strategy parameter set, the candidate strategy parameter set including a look-ahead distance adjustment amount and a speed adjustment amount, wherein the look-ahead distance adjustment amount is used to represent the distance of the target point in front of the path referred to by the robot during travel, and the speed adjustment amount is used to represent the change amplitude of the travel speed of the robot under the current traffic condition.

[0064] The candidate strategy generation unit divides the deviation score result into different intervals, and correspondingly generates multiple candidate parameter sets of the look-ahead distance adjustment amount and the speed adjustment amount, so that the system has adaptive response capability for different deviation degrees.

[0065] The optimal strategy selection unit compares the candidate strategy parameter set according to the deviation score result and the passability factor.

[0066] The initial trajectory deviation difference before and after the change of the look-ahead distance adjustment amount is calculated, and the initial trajectory deviation difference is taken as the deviation correction effect data.

[0067] The difference of the stability feature data before and after the change of the speed adjustment amount is calculated, and the difference of the stability feature data is taken as the passing stability influence data.

[0068] The deviation correction effect data and the passing stability influence data are compared. When the deviation correction effect data is greater than the change amount of the passing stability influence data, the strategy parameter with the look-ahead distance adjustment amount greater than the preset look-ahead distance threshold is selected.

[0069] When the passing stability influence data is greater than the change amount of the deviation correction effect data, the strategy parameter with the speed adjustment amount greater than the preset speed adjustment amount threshold is selected.

[0070] The selected strategy parameter is output as the optimal strategy parameter.

[0071] The optimal strategy selection unit calculates the difference in trajectory deviation before and after the adjustment of the look-ahead distance adjustment amount and the difference in stability feature before and after the adjustment of the speed adjustment amount, compares the change amounts, selects a parameter group with better deviation correction effect, outputs the optimal strategy parameter, and realizes intelligent decision-making based on quantized data.

[0072] The execution control unit is configured to adjust the wheel speed difference and the forward direction according to the optimal strategy parameter, and output execution control data.

[0073] The execution control unit generates wheel speed difference and forward direction control instructions according to the optimal strategy parameter, and performs real-time smooth adjustment on the robot motion, ensuring continuous and smooth changes in steering and speed, effectively improving path keeping stability and orchard passage safety.

[0074] The tracking strategy generation module realizes the conversion of path deviation quantization results to specific control instructions, and is a core link for the system to realize autonomous path correction and stable driving. Through candidate parameter generation, data-driven optimal strategy selection, and execution control closed loop, the module can dynamically adjust the look-ahead distance and speed according to the deviation score and terrain conditions, ensure the robot to maintain path accuracy and stable posture in different orchard passage environments, and realize efficient and continuous path tracking control.

[0075] The dynamic channel regulation module includes a meeting identification unit, a meeting regulation unit, and a headland turning guidance unit.

[0076] The meeting identification unit reads the position, direction of travel, and speed data of the current robot and other robots from the alignment data set, calculates the relative distance according to the position difference, calculates the relative speed according to the speed difference, and forms the time-to-collision data through the ratio of the relative distance and the relative speed. The time-to-collision data is compared with the channel geometric constraint data formed by the effective width of the channel and the tree canopy obstruction to determine whether both can pass at the same time under the current channel conditions, and meeting identification data is generated.

[0077] The meeting identification unit calculates the relative distance and relative speed by comparing the position, direction, and speed data of multiple robots, and generates meeting identification data by combining the time-to-collision index and channel geometric constraint information, so as to accurately determine whether there is a potential collision conflict and its urgency.

[0078] The meeting regulation unit is configured to select a waiting, insertion, or yielding strategy according to the meeting identification data and the deviation score result. After judging the position relationship between the robot and other robots, the passable factor, and the comprehensive result of the path deviation score: If the meeting identification data indicates a potential collision risk and the deviation score is lower than the preset deviation score threshold, the waiting strategy is selected.

[0079] If it is necessary to form a passing order in the multi-robot flow, an insertion strategy is selected.

[0080] When the deviation score is higher than the preset deviation score threshold and the channel condition allows, a yielding strategy is selected.

[0081] When the meeting conflict is judged to be likely to increase, the path merging strategy is preferentially selected.

[0082] The meeting regulation unit selects a strategy based on the meeting identification data and the deviation score result, generates a waiting, insertion or yielding instruction automatically in the multi-robot operation through quantitative comparison of the collision risk, the deviation degree and the channel passability, and realizes dynamic coordination and safe avoidance of the orchard channel traffic flow.

[0083] The end-of-path turning guidance unit is used to perform a path switching action at the end of the path, and the path switching action is: A transition posture is generated in advance before approaching the end of the path, a pre-swing adjustment is performed to align with the initial posture of turning, a turning action is completed according to the set curvature of the curve by controlling the wheel speed difference, and the robot is realigned to the center axis of the next reference path through the merging process.

[0084] The end-of-path turning guidance unit performs continuous posture adjustment of pre-swing, turning and merging at the end of the path, and through accurate calculation of the curvature control and wheel speed difference distribution, the robot smoothly completes path connection, avoids collision of fruit trees or shaking of goods due to sudden turning or posture mutation, and ensures the coherence and stability of the operation process.

[0085] The dynamic channel regulation module realizes intelligent avoidance and path switching control in the multi-robot operation environment of the orchard, so that the robot maintains operation continuity and safety in narrow channels, meetings and end-of-path areas. Through meeting identification, strategy regulation and end-of-path turning guidance, the module can real-time perceive the channel resource occupation state, predict potential conflicts and autonomously generate yielding or path switching instructions, and ensure traffic order and smooth path connection in the multi-robot cooperation process.

[0086] The feedback correction module includes a feedback data acquisition unit, a threshold updating unit and a weight adjustment unit.

[0087] The feedback data acquisition unit is used to acquire trajectory deviation, posture fluctuation and through stability data generated in the robot operation process after the path tracking strategy is executed, record the trajectory deviation, posture fluctuation and through stability data in time sequence and perform effectiveness check, and generate feedback data.

[0088] The feedback data acquisition unit collects trajectory deviation, attitude fluctuation and traffic stability data during the operation of the robot after the path tracking is completed, and performs time sequence recording and effectiveness verification to generate feedback data that can truly reflect the operation state and provide a reliable basis for subsequent correction of the system.

[0089] The threshold updating unit analyzes the change trend of trajectory deviation and attitude fluctuation in the feedback data, and determines whether the current deviation score threshold can accurately reflect the actual operation state of the robot. When the feedback data indicates that there is insufficient response, a deviation score threshold adjustment signal is output.

[0090] The threshold updating unit further combines execution control data and time difference for lag compensation when updating the deviation score threshold.

[0091] The threshold updating unit analyzes the trajectory deviation and attitude fluctuation trend in the feedback data to determine the consistency of the deviation score threshold and the actual operation performance. When response delay or insufficient sensitivity is detected, the threshold is corrected in combination with the execution control data and time lag, ensuring that the scoring mechanism can reflect the current state of the robot in real time.

[0092] The weight adjustment unit compares and analyzes the influence of different feature changes in the feedback data on the operation effect, determines the feature influence that should be enhanced or weakened in the subsequent deviation prediction and scoring process, and outputs a set of correction parameters.

[0093] The weight adjustment unit compares and analyzes the relevance of different feature changes in the feedback data and the operation effect, determines the feature weight that should be enhanced or weakened in the subsequent deviation prediction and scoring process, and outputs a set of correction parameters, so that the system gradually forms adaptive optimization ability in continuous operation, improving path tracking accuracy and overall stability.

[0094] The feedback correction module realizes adaptive optimization and performance backfilling after system operation, so that the orchard transport robot has the ability of continuous learning and self-correction. This module records and analyzes the trajectory deviation, attitude fluctuation and traffic stability data during the path tracking execution process, dynamically updates the scoring threshold and feature weight, thereby correcting the sensitivity of the deviation prediction and strategy generation model, realizing the self-evolution and long-term stable optimization of the control parameters.

[0095] Those skilled in the art will appreciate that embodiments of the present application can be readily used as a method, a system or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (or computer- readable storage media) having computer-usable program code embodied in the medium. The medium can be any available medium or combination thereof that is accessible by a general purpose or special purpose computer. By way of example, such computer-usable storage media can include a volatile memory, such as a random access memory (RAM), a non-volatile memory, such as a read-only memory (ROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a floppy diskette, a compact disk, a hard disk, or any other medium that can be used to carry or store computer-usable program code in the form of computer-usable instructions or data structures and that can be accessed by a general purpose or special purpose computer, or a general-purpose or special-purpose processor. Also, the present application can be embodied in a computer program product which can be executed in particular by a general purpose or special purpose computer or a general-purpose or special-purpose processor. The computer program product can comprise a computer-readable storage medium, such as the memory (RAM), the disc (CD-ROM or DVD), or the hard disk, having computer-usable program code embodied in the medium. The computer-usable program code can cause a computer, and in particular the processor, to carry out one or more methods or parts of methods according to the present application. The computer program product can have been transferred from the place where the computer program has been developed to the place where it is to be used (for example, from a developer's Figure 1 one or more flows and / or blocks Figure 1 one or more flows and / or blocks

[0096] It should be noted that the above-mentioned embodiments are only used to illustrate but not to limit the technical solutions of the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalent replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A path tracking system for orchard transport robots, characterized in that, It includes an environmental acquisition and alignment module, a travel feature analysis module, a deviation prediction and scoring module, a tracking strategy generation module, a dynamic channel control module, and a feedback correction module; The environmental acquisition and alignment module is used to collect the position information and environmental structure information of the robot when it runs between rows in the orchard, and to perform synchronization processing based on the master control clock to generate an alignment dataset. The travel feature analysis module is used to extract stability features, terrain navigability features and initial trajectory deviation data based on the aligned dataset; The deviation prediction and scoring module is used to generate deviation scoring results based on initial trajectory deviation data and drivability features; The tracking strategy generation module is used to generate candidate strategy parameters based on the deviation scoring results, select the optimal strategy, and execute control. The dynamic channel control module is used to perform obstacle avoidance and path switching in scenarios where multiple robots are running simultaneously in an orchard and at the end of a path. The feedback correction module is used to collect execution feedback data and update the scoring threshold and parameter weights.

2. The orchard transport robot path tracking system as described in claim 1, characterized in that, The environmental acquisition and alignment module includes a position and attitude observation unit, a row and column structure recognition unit, and a time alignment unit; The position and attitude observation unit is used to collect satellite positioning data, heading angle data and attitude angle data, and to perform synchronous acquisition and validity verification of the satellite positioning data, heading angle data and attitude angle data, and output position and attitude observation data. The row and column structure recognition unit is used to acquire external scene information and identify the direction between fruit tree rows, effective width of the channel, and position of the left and right boundaries based on the arrangement characteristics of the fruit tree planting rows. Collect information on the outer edge outline of the fruit tree crown, the arrangement of the trunk, and the reflection of the ground texture. Extract spatial features and perform geometric morphology analysis on the collected scene data. Based on the changes in crown density, the trend of row offset, and the bending characteristics between rows, determine whether there is local narrowing, turning, or offset in the direction of the fruit tree rows and columns, and output the row and column structure data. The time alignment unit is used to perform time synchronization processing on position and attitude observation data and row and column structure data with the master control clock as a reference, and the output is an aligned dataset.

3. The orchard transport robot path tracking system as described in claim 2, characterized in that, The travel feature analysis module includes a stability feature extraction unit, a terrain navigability analysis unit, and a baseline trajectory deviation calculation unit. The stability feature extraction unit is used to analyze the robot's running stability in the fruit tree row passage based on the aligned dataset. By jointly judging the heading angle change curve, attitude angle fluctuation range and travel speed continuity, it identifies whether the robot has a yaw tendency, lateral offset trend or decreased travel stability. The output is a change feature. The change feature is smoothed, periodically analyzed and trend extracted to output stability feature data. The terrain accessibility analysis unit extracts surface feature changes and attitude features from the aligned dataset and outputs accessibility factor data. The baseline trajectory deviation calculation unit calculates the lateral position difference and the heading direction difference based on the correspondence between satellite positioning data and the reference path center axis in the same coordinate system, and outputs the initial trajectory deviation data. The initial trajectory deviation data is used to output a numerical form representing the degree of deviation of the robot's current position from the ideal travel path; The reference path center axis is determined based on the channel centerline obtained by the row and column structure identification unit and the operation path planning model.

4. The orchard transport robot path tracking system as described in claim 3, characterized in that, The deviation prediction and scoring module includes a trajectory deviation trend prediction unit, a drivability correction unit, and a deviation scoring unit. The trajectory deviation trend prediction unit is used to analyze the trend of trajectory deviation over time based on initial trajectory deviation data and stability characteristic data. By jointly comparing the lateral offset, heading angle offset and attitude angle change rate, it identifies whether the deviation is aggravated, alleviated or stabilized, and outputs predicted deviation data. The trajectory deviation trend prediction unit smooths out instantaneous amplitude abrupt changes during the analysis process and performs trend fitting on continuous deviation segments. The mobility correction unit is used to correct the prediction deviation data based on mobility factor data. The correction process involves associating the prediction deviation data with the mobility factor data. When the mobility factor indicates that the terrain resistance is large or the instability is increasing, the sensitivity to the change in prediction deviation is increased to a first sensitivity value, and the corrected deviation data is output. The deviation scoring unit is used to perform weighted summation based on the corrected deviation data to generate a deviation scoring result.

5. The orchard transport robot path tracking system as described in claim 4, characterized in that, The tracking strategy generation module includes a candidate strategy generation unit, an optimal strategy selection unit, and an execution control unit; The candidate strategy generation unit is used to divide the deviation score results into intervals based on the deviation score results and generate a candidate strategy parameter set. The candidate strategy parameter set includes forward look distance adjustment and speed adjustment. The forward look distance adjustment is used to characterize the distance of the target point ahead of the path referenced by the robot during the movement, and the speed adjustment is used to characterize the magnitude of the change in the robot's movement speed under the current traffic conditions. The optimal strategy selection unit compares the candidate strategy parameter set based on the deviation score result and the feasibility factor. Calculate the difference in initial trajectory deviation before and after the change in forward sight distance adjustment, and use the difference in initial trajectory deviation as the deviation correction effect data; The difference in stability characteristic data before and after the change in speed adjustment is calculated, and the difference in stability characteristic data is used as the data affecting stability. Compare the bias correction effect data with the stability effect data: When the deviation correction effect data is greater than the change in the stability-affected data, select a strategy parameter where the forward look distance adjustment is greater than the preset forward look distance threshold. When the change in data due to stability effects is greater than the change in data due to deviation correction, select a strategy parameter whose speed adjustment amount is greater than the preset speed adjustment amount threshold. The selected strategy parameters are output as the optimal strategy parameters; The execution control unit is used to adjust the wheel speed difference and forward direction according to the optimal strategy parameters, and output execution control data.

6. The orchard transport robot path tracking system as described in claim 5, characterized in that, The dynamic channel control module includes a meeting vehicle recognition unit, a meeting vehicle control unit, and a turning guidance unit. The vehicle meeting recognition unit reads the position, direction of travel, and speed data of the current robot and other robots from the alignment dataset, calculates the relative distance based on the position difference, calculates the relative speed based on the speed difference, and forms time-to-collision data by the ratio of the relative distance to the relative speed. The time-to-collision data is compared with the effective width of the channel and the channel geometric constraint data formed by the canopy occlusion to determine whether the two vehicles can pass at the same time under the current channel conditions, and generates vehicle meeting recognition data. The vehicle meeting control unit is used to select a waiting, insertion, or yielding strategy based on the vehicle meeting recognition data and deviation scoring results: After considering the robot's positional relationship with other robots, accessibility factors, and path deviation scores: If the oncoming traffic recognition data indicates a potential collision risk and the deviation score is lower than the preset deviation score threshold, the waiting strategy is selected. If a passage order needs to be established in a multi-robot flow, select an insertion strategy; When the deviation score is higher than the preset deviation score threshold and the channel conditions allow, the yielding strategy is selected; The heading guidance unit is used to perform a path switching action at the end of the path, and the path switching action is as follows: Before approaching the end of the path, a transition posture is generated in advance, and the robot is pre-swinged to align with the initial steering posture. The steering action is completed by controlling the wheel speed difference according to the set curve curvature. The robot is then realigned to the center axis of the next reference path through the lane merging process.

7. The orchard transport robot path tracking system as described in claim 6, characterized in that, The feedback correction module includes a feedback data acquisition unit, a threshold update unit, and a weight adjustment unit; The feedback data acquisition unit is used to collect trajectory deviations, attitude fluctuations, and stability data generated during the robot's operation after the path tracking strategy is executed. It performs time-series recording and validity checks on the trajectory deviations, attitude fluctuations, and stability data to generate feedback data. The threshold update unit analyzes the changing trends of trajectory deviation and attitude fluctuation in the feedback data, determines whether the current deviation score threshold can accurately reflect the actual operating state of the robot, and outputs a deviation score threshold adjustment signal when the feedback data indicates that there is insufficient response. The weight adjustment unit compares and analyzes the impact of different feature changes in the feedback data on the running effect, determines the feature influence that should be strengthened or weakened in the subsequent deviation prediction and scoring process, and outputs a set of correction parameters.

8. The orchard transport robot path tracking system as described in claim 7, characterized in that, The deviation scoring unit further includes a dynamic weight adjustment mechanism; The dynamic weight adjustment mechanism is used to adjust the weight ratio of stability characteristics and accessibility factors in deviation scoring in real time according to the changing trends of different orchard terrains and operating conditions.

9. The orchard transport robot path tracking system as described in claim 8, characterized in that, When the vehicle passing control unit determines that the vehicle passing conflict may continue to increase, it prioritizes the path merging strategy.

10. The orchard transport robot path tracking system as described in claim 9, characterized in that, The threshold update unit further incorporates execution control data and time difference for lag compensation when updating the deviation score threshold.

Images