Multi-sensor fusion agricultural power machine automatic obstacle avoidance control method and system

Abstract

The invention discloses a multi-sensor fusion agricultural power machine automatic obstacle avoidance control method and system, and relates to the technical field of intelligent control. The method comprises the steps of performing farmland range detection scanning under a preset range constraint based on a sensing array along with driving of an agricultural machine, and obtaining multi-source sensing data; by uploading agricultural tasks and combining a predefined classification mode of farmland obstacles and a farmland topological graph, a rapid judgment device and an autonomous obstacle avoidance module are developed in an agricultural machine center controller; dynamic weighting fusion is carried out on multi-source sensing data, local obstacle verification based on a rapid judger and passable area decision and obstacle avoidance optimization decision based on an autonomous obstacle avoidance module are triggered, and an obstacle avoidance strategy is generated; and an execution unit driven by the agricultural machinery responds to the obstacle avoidance strategy and executes automatic obstacle avoidance management. The technical problems that in the prior art, obstacle recognition response of agricultural machinery is delayed in a complex farmland environment, and obstacle avoidance behavior decision is unstable are solved, and the technical effect of improving the automatic obstacle avoidance reliability of the agricultural machinery is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control, in particular to a multi-sensor fusion agricultural power machine automatic obstacle avoidance control method and system. BACKGROUND

[0002] In the process of agricultural machinery automation operation, the agricultural machinery needs to realize autonomous navigation in the unstructured, complex terrain and dynamically changing environment of farmland scene. The existing agricultural machinery obstacle avoidance technology mostly relies on single sensor or static threshold judgment method, which is difficult to consider multiple source environment characteristics at the same time, resulting in delayed obstacle identification response, high false detection and missed detection rate in the case of light change, dust environment, crop shielding or limited field of view. In addition, the traditional obstacle avoidance strategy is usually based on fixed rules or simple path offset, which lacks dynamic analysis ability of passable area, so that the agricultural machinery is prone to unstable decision, obstacle avoidance failure or stagnation when facing multiple obstacle combinations, complex terrain or narrow working space, which cannot meet the continuous, efficient and reliable agricultural automation operation demand. SUMMARY

[0003] The present application provides a multi-sensor fusion agricultural power machine automatic obstacle avoidance control method and system, which solves the technical problems of delayed obstacle identification response and unstable obstacle avoidance behavior decision of agricultural machinery in complex farmland environment in the prior art.

[0004] The first aspect of the present application provides a multi-sensor fusion agricultural power machine automatic obstacle avoidance control method, which comprises: Driving the agricultural machinery, based on the sensor array loaded on the agricultural machinery, performing farmland range detection scanning under preset range constraint to obtain multi-source sensing data; uploading agricultural tasks, combining the pre-defined classification mode of farmland obstacles and the farmland topological map, developing a rapid judge and autonomous obstacle avoidance module in the agricultural machinery central control; dynamically weighting and fusing the multi-source sensing data to trigger local obstacle detection based on the rapid judge, and passable area decision and obstacle avoidance optimization decision based on the autonomous obstacle avoidance module, to generate an obstacle avoidance strategy; the execution unit driven by the agricultural machinery mechanical drive responds to the obstacle avoidance strategy to execute automatic obstacle avoidance management.

[0005] The second aspect of the present application provides a multi-sensor fusion agricultural power machine automatic obstacle avoidance control system, which comprises: The data acquisition component: driven by the agricultural machine, based on the sensing array loaded by the agricultural machine, performing farmland range detection scanning under preset range constraints to obtain multi-source sensing data; the judgment development component: through uploading an agricultural task, combining a predefined classification mode of farmland obstacles and a farmland topological map, developing a rapid judgment device and an autonomous obstacle avoidance module in the agricultural machine central control; the strategy generation component: through dynamic weighting fusion of the multi-source sensing data, triggering local obstacle detection based on the rapid judgment device, and passable area decision and obstacle avoidance optimization decision based on the autonomous obstacle avoidance module, generating an obstacle avoidance strategy; the execution management component: an execution unit driven by the agricultural machine mechanically responds to the obstacle avoidance strategy to perform automatic obstacle avoidance management.

[0006] One or more technical solutions provided in the present application have at least the following technical effects or advantages: First, the data acquisition component: driven by the agricultural machine, based on the sensing array loaded by the agricultural machine, performing farmland range detection scanning under preset range constraints to obtain multi-source sensing data. Next, the judgment development component: through uploading an agricultural task, combining a predefined classification mode of farmland obstacles and a farmland topological map, developing a rapid judgment device and an autonomous obstacle avoidance module in the agricultural machine central control. Then, the strategy generation component: through dynamic weighting fusion of the multi-source sensing data, triggering local obstacle detection based on the rapid judgment device, and passable area decision and obstacle avoidance optimization decision based on the autonomous obstacle avoidance module, generating an obstacle avoidance strategy. Finally, the execution management component: an execution unit driven by the agricultural machine mechanically responds to the obstacle avoidance strategy to perform automatic obstacle avoidance management. The technical problems of obstacle recognition response lag and unstable obstacle avoidance behavior decision of the agricultural machine in the complex farmland environment in the prior art are solved, and the technical effect of improving the reliability of automatic obstacle avoidance of the agricultural machine is achieved. BRIEF DESCRIPTION OF DRAWINGS

[0007] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0008] Figure 1 The flow chart of the automatic obstacle avoidance control method of the agricultural power machine based on multi-sensor fusion provided by the embodiments of the present application; Figure 2 The structure schematic diagram of the automatic obstacle avoidance control system of the agricultural power machine based on multi-sensor fusion provided by the embodiments of the present application.

[0009] Explanation of reference signs: data acquisition component 11, judgment development component 12, strategy generation component 13, execution management component 14. DETAILED DESCRIPTION

[0010] To further illustrate the technical means and effects adopted by the present application to achieve the predetermined inventive purpose, the specific embodiments, structures, features and effects according to the present application are described in detail below in combination with the drawings and preferred embodiments.

[0011] In one embodiment, as shown in the accompanying drawings, the present application provides a multi-sensor fusion agricultural power machine automatic obstacle avoidance control method, wherein the method comprises: Figure 1 Based on the sensor array loaded on the agricultural machine, the farmland range detection scanning under the preset range constraint is performed, and multi-source sensing data is obtained.

[0012] During the driving process of the agricultural machine, the central control of the agricultural machine continuously schedules the sensor array loaded on the front end, sides and top of the machine body to detect and scan the farmland environment. The sensor array includes at least one or more of a laser radar, a millimeter wave radar, a depth camera, an ultrasonic sensor and a soil state sensor, and can periodically output distance information, reflection intensity information, image information, depth information and soil hardness and moisture parameters within the preset detection range around the driving direction of the agricultural machine. The central control of the agricultural machine sets the detection range constraint according to the operation requirement, limits the sensor array to perform scanning within the specified forward distance range, lateral coverage width and vertical detection angle, and adjusts the sampling frequency and scanning period of each sensor in real time according to the driving speed of the agricultural machine, so that the detection area is continuously covered with the movement of the agricultural machine. The central control system performs time synchronization, coordinate calibration and noise preprocessing on the raw data reported by each sensor in each period, constructs a unified farmland environment perception data format, and finally forms multi-source sensing data including surface morphology, obstacle contour, crop height, crop density, soil condition and dynamic object features, which provides basic input for subsequent obstacle identification and obstacle avoidance decision.

[0013] By uploading an agricultural task, combining a pre-defined farmland obstacle classification mode and a farmland topological map, a rapid judge and autonomous obstacle avoidance module are developed in the central control of the agricultural machine.

[0014] ​Before agricultural machinery operations begin, the user or the higher-level agricultural scheduling system uploads the corresponding agricultural task to the agricultural machinery central control unit via a communication interface. This task includes elements such as operation type, target area, operation sequence requirements, prohibited areas, and safety constraints. Upon receiving the task, the agricultural machinery central control unit first calls a predefined farmland obstacle classification model, categorizing obstacles into hard obstacles, soft obstacles, traversable objects, and operation targets to guide subsequent environmental judgment logic. Simultaneously, the system loads the corresponding farmland topology map, which consists of farmland structure information such as the operation target area, machinery travel path, and adjacent areas, describing the spatial distribution of the farmland. Based on the agricultural task content, obstacle classification model, and farmland topology map, the agricultural machinery central control unit constructs an environmental semantic data structure associated with the task, and develops two functional modules on this basis: a rapid judgment unit for quickly classifying and determining the effectiveness of potential obstacles in the local environment; and an autonomous obstacle avoidance module for performing traversable area analysis and obstacle avoidance optimization decisions according to a predetermined strategy after detecting an obstacle. Both the rapid judgment unit and the autonomous obstacle avoidance module take task constraints as input and are invoked in real time according to the operation cycle of the agricultural machinery central control unit, providing joint support at the task level and space level for the subsequent generation of obstacle avoidance strategies.

[0015] Furthermore, an agricultural prior knowledge definition classification model is implemented, in which farmland areas are classified into hard obstacles, soft obstacles, traversable objects, and work objects; through agricultural task interpretation, task elements are determined, wherein the task elements include a task priority matrix and a set of constraints; for farmland areas, the work object area, the machinery travel path, and adjacent areas are identified as farmland topology maps.

[0016] Before performing agricultural tasks, the agricultural machinery control system first defines prior agricultural knowledge to construct a semantic classification model for farmland obstacles and the working environment. This classification model, based on common physical components in agricultural production activities, divides farmland areas into four categories: hard obstacles, soft obstacles, traversable objects, and work objects. Hard obstacles include fixed entities that cannot be passed, such as rocks, tree stumps, and farmland facilities; soft obstacles include deformable elements that must be avoided, such as lodged crops and surface stubble; traversable objects include low-lying weeds and lightweight, compressible areas; and work objects include crop planting zones and areas with plants awaiting treatment, areas that must be strictly protected.

[0017] After loading agricultural tasks, the agricultural machinery central control system interprets and analyzes the task content, extracting task elements, including a task priority matrix and a set of constraints. The task priority matrix represents the priority order among various task objectives during operation, such as ensuring crop safety, maintaining operational efficiency, and avoiding critical obstacles. The set of constraints defines prohibited areas, permissible deviation ranges, and safe distance thresholds during operation to guide subsequent environmental assessments and path decisions.

[0018] Based on the aforementioned prior agricultural knowledge and task elements, the agricultural machinery control system further performs spatial structure analysis on the target farmland area, marking areas including planting rows, work boundaries, regular driving routes, and buffer zones. This identifies the work target area, the machinery's travel path, and its adjacent areas, and constructs a farmland topology map to describe the structural relationships of the farmland environment. This farmland topology map serves as the basic input for subsequent rapid judgment and obstacle avoidance modules, supporting local obstacle detection, traversable area analysis, and obstacle avoidance strategy generation.

[0019] By dynamically weighting and fusing the multi-source sensor data, a local obstacle detection based on a fast judge is triggered, and an obstacle avoidance strategy is generated based on the passable area decision and obstacle avoidance optimization decision based on the autonomous obstacle avoidance module.

[0020] The agricultural machinery central control system dynamically weights and fuses multi-source sensor data. First, it assigns dynamic weight coefficients to various sensor data based on the effectiveness of different sensors in the current farmland environment, combined with their historical stability, noise level, and environmental adaptability. The central control system then uses these dynamic weights to fuse multi-source data, including lidar point clouds, camera images, millimeter-wave echoes, and ultrasonic ranging, generating a real-time comprehensive environmental matrix that characterizes terrain undulations, obstacle morphology, crop structure, and soil conditions. Subsequently, the central control system calls its embedded fast judgment unit, using the real-time environmental matrix as input, to quickly detect obstacle types, boundary locations, and passage risks within a local area, obtaining local obstacle detection results. After obtaining the local obstacle detection results, the agricultural machinery central control unit further activates the autonomous obstacle avoidance module. Based on the passable area decision logic, combined with the updated farmland topology map and real-time perception matrix, it determines the effective passable area under the current operating conditions. Then, based on the obstacle avoidance optimization decision logic, according to preset path cost constraints and task priorities, it generates multiple candidate obstacle avoidance paths within the passable area and evaluates the cost of each path to determine the optimal obstacle avoidance scheme that meets both safety and operational continuity requirements. Finally, the central control unit comprehensively generates an obstacle avoidance strategy based on the local obstacle detection results, passable area decision results, and obstacle avoidance optimization decision results. This strategy is then used as the control input for subsequent mechanical execution units to achieve the agricultural machinery's automatic obstacle avoidance behavior.

[0021] Furthermore, the dynamic weighting and fusion of the multi-source sensor data includes: The system receives the multi-source sensor data, performs dynamic weighted fusion of the multi-source sensors in the context of the farmland environment, and determines a real-time perception matrix. Based on the real-time perception matrix, it performs real-time local updates of physical elements in the farmland topology map.

[0022] The agricultural machinery central control system receives multi-source sensor data from LiDAR, cameras, millimeter-wave radar, ultrasonic sensors, and soil condition sensors during each environmental perception cycle. The system evaluates the effectiveness of each sensor based on current farmland environmental conditions (such as light intensity, dust concentration, crop shading, and surface humidity), assigning corresponding weight coefficients to different sensor data through a preset dynamic weight allocation mechanism. Subsequently, the multi-source data is synchronized, filtered, and matched according to a unified spatial coordinate system and timestamps, fusing them to form a real-time perception matrix that characterizes farmland topography, crop structure, obstacle boundaries, and soil properties. Based on this real-time perception matrix, the central control system maps the feature information reflecting the status of the work object, path accessibility, obstacle outlines, and surface conditions to corresponding regional nodes on the farmland topology map, performing real-time local updates of the physical elements in the farmland topology map. The updates include dynamic corrections to crop area status, updates to the marking of potential obstacle locations and attributes, and adjustments to the boundaries of passable areas. These updates ensure that the farmland topology map maintains a local structure consistent with the actual environment, providing accurate basic data support for obstacle detection by the rapid judgment device and path decision-making by the autonomous obstacle avoidance module.

[0023] Furthermore, the generation of obstacle avoidance strategies includes: The agricultural machinery central control unit performs local obstacle detection on the updated farmland topology map based on the embedded fast judgment device and determines the judgment result; based on the judgment result, it generates obstacle avoidance instructions; based on the obstacle avoidance instructions, it activates the autonomous obstacle avoidance module embedded in the agricultural machinery central control unit, executes two-stage obstacle avoidance decisions, and generates an obstacle avoidance strategy.

[0024] After updating the real-time perception matrix, the agricultural machinery central control unit calls the embedded fast judgment unit to perform obstacle detection on the local areas related to the movement of the agricultural machinery in the updated farmland topology map. Based on the physical element categories, obstacle boundary positions, and attribute information identified in the topology map, combined with environmental features in the real-time perception matrix, the fast judgment unit quickly classifies and judges whether there are hard obstacles, soft obstacles, traversable objects, or work objects in the current direction of travel and adjacent areas, and outputs the corresponding judgment results. When the judgment result indicates the presence of an obstacle that needs to be avoided, the agricultural machinery central control unit generates a preliminary obstacle avoidance command based on the obstacle category, relative position of the obstacle, and work task constraints. The command content includes avoidance requirements, passable area requirements, path offset direction, and safety distance requirements.

[0025] After generating the obstacle avoidance command, the central control system activates the embedded autonomous obstacle avoidance module and performs obstacle avoidance analysis according to a two-stage obstacle avoidance decision-making process: The first decision layer analyzes the current farmland topology map and real-time perception matrix based on the obstacle avoidance command to determine the passable area that meets operational constraints and safety conditions; the second decision layer performs obstacle avoidance optimization within the passable area, determining the optimal obstacle avoidance path through operations such as path search, avoidance magnitude adjustment, and path cost evaluation. Finally, the autonomous obstacle avoidance module integrates the output results of the two-stage decisions to generate a complete obstacle avoidance strategy and submits this strategy to the agricultural machinery central control system to drive subsequent execution units for automatic obstacle avoidance control.

[0026] Furthermore, the autonomous obstacle avoidance module includes a first decision layer and a second decision layer. The first decision layer performs decisions on passable areas, and the second decision layer performs obstacle avoidance optimization decisions. The second decision layer has a built-in progressive obstacle avoidance mechanism and path cost function.

[0027] The autonomous obstacle avoidance module comprises a first decision layer and a second decision layer, responsible for making decisions throughout the entire process, from environmental analysis to generating the optimal obstacle avoidance path. The first decision layer executes the passable area decision. Based on obstacle avoidance instructions, the updated farmland topology map, and the real-time perception matrix, it performs regional-level filtering analysis of the current operational scenario. Under the premise of meeting operational priorities and safety constraints, it defines the spatial range within which the agricultural machinery can safely pass at the current moment and dynamically corrects the boundaries of the passable area. The second decision layer executes the obstacle avoidance optimization decision. Using the passable area output by the first decision layer as the search constraint, it generates the optimal path scheme that meets the obstacle avoidance requirements through path search, path evaluation, and path correction processes. The second decision layer integrates a progressive obstacle avoidance mechanism. Through multi-level strategies such as path fine-tuning, speed adjustment, local replanning, and emergency stopping when necessary, the obstacle avoidance scheme has the ability to continuously adjust from light avoidance to forced avoidance. Meanwhile, the second decision layer also has a built-in path cost function, which evaluates different candidate paths based on factors such as crop damage risk, operational efficiency, energy consumption, and soil compaction degree. This function is used to screen the obstacle avoidance path with the lowest cost and the least operational impact, ensuring that the generated obstacle avoidance strategy takes into account safety, feasibility, and operational continuity.

[0028] Furthermore, the first decision-making layer executes decisions regarding accessible areas, including: Based on the updated farmland topology map, a passable area is defined; wherein, the method for defining the passable area includes: determining the crop status based on the real-time sensing matrix and assessing the agronomically permissible passable area; determining the soil conditions based on the real-time sensing matrix and assessing the actual safe passable area; determining the physical parameters of agricultural machinery based on the real-time sensing matrix and calculating the theoretically passable area; and defining the passable area by taking the intersection of the agronomically permissible passable area, the actual safe passable area, and the theoretically passable area.

[0029] After acquiring the updated farmland topology map, the agricultural machinery central control system uses the marked work object areas, obstacle areas, and passageways on the topology map as a basis, combined with environmental features from the real-time perception matrix, to perform a passability analysis of the current work space and define passable areas. The definition of passable areas includes: First, based on crop status information such as crop height, density, and structural integrity reflected in the real-time perception matrix, assessing the compressibility, damage tolerance, and agronomical protection needs of the crops within the area to determine the agronomically permissible passage area; second, based on soil conditions such as soil moisture, surface hardness, and ground continuity from the real-time perception matrix, identifying the existence of risks such as getting stuck, slipping, or insufficient load-bearing capacity to determine the actual safe passage area; third, based on agricultural machinery physical parameters extracted from the real-time perception matrix, such as the machinery's posture, body size, wheelbase, turning radius, and minimum passage width, calculating the theoretically achievable passage area under the current machine constraints. Ultimately, the first decision-making level determined the geometric intersection of the agronomically permissible passage area, the practically safe passage area, and the theoretically passable area, and identified the area that collectively met the constraints of agronomic protection, safety conditions, and equipment capabilities as the final passable area.

[0030] Furthermore, the path cost function is a weighted sum of crop damage cost, operational efficiency cost, energy consumption cost, and soil compaction cost; wherein, the crop damage cost is defined by the compaction area and damage recoverability, the operational efficiency cost is defined by the detour distance, speed loss, and impact on subsequent operations, the energy consumption cost is defined by the path curvature, ground resistance, and speed change, and the soil compaction cost is defined based on the number of passes, soil moisture, and load weight.

[0031] The path cost function is a weighted sum of crop damage cost, operational efficiency cost, energy consumption cost, and soil compaction cost, used to quantitatively evaluate candidate obstacle avoidance paths. By assigning corresponding weight coefficients to different cost items, the path cost function ensures that the obstacle avoidance optimization process takes into account crop protection, operational efficiency, energy consumption, and soil environmental impact, thereby ensuring that the generated obstacle avoidance strategy meets the comprehensive performance requirements of agricultural operations. The crop damage cost is defined by the area of ​​crop damage that the path may cause in the farmland and the recoverability of the crop damage: the larger the area of ​​crop damage or the higher the irrecoverability of the damage, the greater the crop damage cost. The operation efficiency cost is defined by the detour distance of the path, the speed loss, and the degree of impact on subsequent operation processes: the greater the path deviation, the greater the speed adjustment range, or the higher the risk of subsequent operation trajectory deviation, the greater the operation efficiency cost. The energy consumption cost is defined by the path curvature, ground resistance, and speed changes: the greater the path curvature, the more turns are required and the higher the power demand; the greater the ground resistance or the more frequent the speed changes, the greater the energy consumption cost. The soil compaction cost is defined based on the number of passes, soil moisture, and load weight: when the same area is passed through multiple times, the soil moisture is high, or the agricultural machinery load is large, the impact on soil compaction is greater, and the soil compaction cost increases accordingly.

[0032] Guided by the path cost function, the autonomous obstacle avoidance module performs multiple rounds of cost evaluation on each candidate path. By weighted summing of the four costs mentioned above, it selects the path with the minimum total cost as the target obstacle avoidance path, providing a quantitative basis for the generation of the final obstacle avoidance strategy.

[0033] Furthermore, the progressive obstacle avoidance machine consists of multiple obstacle avoidance modes, including path fine-tuning, speed adjustment, local replanning, and emergency stop. Obstacle avoidance boundaries are defined using the passable area, and multiple obstacle avoidance modes are determined by segmenting the path of local obstacles. For the first segment of the multi-segment obstacle avoidance mode, multiple rounds of iterative optimization are performed under the constraints of the path cost function and the first obstacle avoidance mode to determine the first obstacle avoidance strategy. Starting from the endpoint of the first obstacle avoidance strategy, obstacle avoidance path analysis is performed on the second segment, until the Nth obstacle avoidance strategy is determined, and the results are then combined to form the obstacle avoidance strategy.

[0034] The progressive obstacle avoidance mechanism consists of multi-level obstacle avoidance modes, including path fine-tuning, speed adjustment, local replanning, and emergency stop, used to progressively increase obstacle avoidance strength according to the risk level of the obstacle. At the beginning of the obstacle avoidance analysis, the autonomous obstacle avoidance module first searches for feasible paths in the current direction of the agricultural machinery and the surrounding area, using the passable area boundary as the obstacle avoidance boundary. When a local obstacle is identified, the autonomous obstacle avoidance module divides the path to be avoided into several segments according to the spatial structure or time sequence of the obstacle, forming a multi-segment obstacle avoidance mode, so that different segments of the path can be optimized independently.

[0035] For the first segment of the multi-segment obstacle avoidance path, the autonomous obstacle avoidance module combines the path cost function with the constraints of the first obstacle avoidance mode (e.g., path fine-tuning or speed adjustment) and performs multiple rounds of iterative optimization. During the iteration process, the path offset, speed change, or passage attitude are continuously corrected based on the evaluation results of the path cost function, thereby determining the first obstacle avoidance strategy that meets the obstacle avoidance requirements. Subsequently, using the endpoint of the first obstacle avoidance strategy as the starting point of the next segment, the autonomous obstacle avoidance module continues to perform obstacle avoidance path analysis on the second segment. By selecting an obstacle avoidance mode suitable for this segment (e.g., local replanning or enhanced path fine-tuning) and calculating the path cost, the second obstacle avoidance strategy is gradually determined. Following this method, the module sequentially completes the determination of obstacle avoidance strategies for segments 1 to N. After all segment optimizations are completed, the obstacle avoidance strategies for each segment are spliced ​​and integrated according to the path temporal sequence or spatial continuity to form a complete obstacle avoidance strategy output, which is used to drive the agricultural machinery to perform obstacle avoidance actions.

[0036] The agricultural machinery drive execution unit responds to the obstacle avoidance strategy and performs automatic obstacle avoidance management.

[0037] After generating an obstacle avoidance strategy, the agricultural machinery central control system parses the strategy into action commands corresponding to the mechanical actuators. These commands include elements such as steering angle adjustment, left and right wheel differential control, drive torque distribution, travel speed correction, and braking control. The central control system sends these action commands to the agricultural machinery drive system, including the steering mechanism, drive wheel controller, hydraulic braking mechanism, and actuator servo units, in real-time control cycles. This ensures that the mechanical actuators act according to the path offset, speed change curve, and safety distance requirements specified in the obstacle avoidance strategy. During the control process, the actuators continuously receive feedback updates from the central control system and dynamically correct the execution process based on the new real-time perception matrix and changes in the local environment. This allows for fine-tuning of the travel direction, gradual adjustment of speed, or stopping when necessary, enabling the agricultural machinery to complete stable obstacle avoidance along the optimal path. The entire execution process is completed within the closed-loop control framework of the central control system, ensuring the continuity, real-time performance, and safety of obstacle avoidance actions in complex farmland environments.

[0038] Furthermore, the agricultural machinery drive's execution unit responds to the obstacle avoidance strategy and performs automatic obstacle avoidance management, including: The obstacle avoidance strategy is decomposed into multiple sub-strategies using the smallest execution unit driven by agricultural machinery. The multiple sub-strategies are then subject to timestamp constraints and marking based on execution unit encoding to generate multi-threaded obstacle avoidance instructions. The agricultural machinery central control unit executes the agricultural machinery obstacle avoidance control by issuing the multi-threaded obstacle avoidance instructions.

[0039] After the obstacle avoidance strategy is generated, the agricultural machinery central control system first decomposes the overall obstacle avoidance strategy into functional sub-strategies based on the structural characteristics and degrees of freedom of the mechanical drive system, according to the smallest execution unit under the agricultural machinery drive. This results in multiple sub-strategies for actions such as steering adjustment, drive control, speed correction, and braking control. Subsequently, the central control system applies time stamp constraints to these sub-strategies to ensure the consistency and safety of their execution sequence. It also adds an execution unit marker to each sub-strategy according to the coding rules of the agricultural machinery's execution units, enabling the system to accurately match the corresponding actuators in multiple execution channels. After completing the time sequencing and execution unit marking of the sub-strategies, the central control system integrates them to generate multi-threaded obstacle avoidance instructions. The multi-threaded structure allows different actions to be executed separately in parallel control channels, avoiding obstacle avoidance instability caused by action conflicts or delays. Finally, the agricultural machinery central control system sends the multi-threaded obstacle avoidance instructions to the execution units of the mechanical drive. The execution units then perform direction adjustments, speed regulation, and necessary braking actions according to the instructions, thereby achieving automatic obstacle avoidance control of the agricultural machinery.

[0040] In summary, the embodiments of this application have at least the following technical effects: First, driven by the agricultural machinery, the sensor array mounted on the machinery performs a field range scan under preset constraints to acquire multi-source sensor data. Next, by uploading agricultural tasks and combining predefined farmland obstacle classification patterns and farmland topology maps, a rapid judgment device and an autonomous obstacle avoidance module are developed within the central control unit of the agricultural machinery. Then, through dynamic weighted fusion of the multi-source sensor data, a local obstacle detection based on the rapid judgment device and a passable area decision and obstacle avoidance optimization decision based on the autonomous obstacle avoidance module are triggered to generate an obstacle avoidance strategy. Finally, the execution unit of the agricultural machinery responds to the obstacle avoidance strategy and performs automatic obstacle avoidance management. This solves the technical problems of slow obstacle recognition response and unstable obstacle avoidance behavior decision-making in complex farmland environments in existing technologies, achieving the technical effect of improving the reliability of automatic obstacle avoidance in agricultural machinery.

[0041] Example 2, based on the same inventive concept as the multi-sensor fusion automatic obstacle avoidance control method for agricultural power machinery in the foregoing examples, such as... Figure 2 As shown, this application provides a multi-sensor fusion-based automatic obstacle avoidance control system for agricultural power machines, wherein the system includes: Data acquisition component 11: Driven by the agricultural machinery, it performs a field range detection and scanning under preset range constraints based on the sensor array mounted on the agricultural machinery to acquire multi-source sensor data; Judgment and development component 12: By uploading agricultural tasks and combining predefined classification patterns of farmland obstacles with farmland topology maps, it develops a fast judgment device and an autonomous obstacle avoidance module within the central control unit of the agricultural machinery; Strategy generation component 13: By dynamically weighting and fusing the multi-source sensor data, it triggers local obstacle detection based on the fast judgment device, and generates an obstacle avoidance strategy based on the passable area decision and obstacle avoidance optimization decision based on the autonomous obstacle avoidance module; Execution management component 14: The execution unit driven by the agricultural machinery responds to the obstacle avoidance strategy and performs automatic obstacle avoidance management.

[0042] Furthermore, the determination development component 12 is used to perform the following method: An agricultural prior knowledge definition classification model is constructed, in which farmland areas are classified into hard obstacles, soft obstacles, traversable objects, and work objects; through agricultural task interpretation, task elements are determined, wherein the task elements include a task priority matrix and a set of constraints; for farmland areas, the work object area, the machinery travel path, and adjacent areas are identified as farmland topology maps.

[0043] Furthermore, the strategy generation component 13 is used to perform the following method: The system receives the multi-source sensor data, performs dynamic weighted fusion of the multi-source sensors in the context of the farmland environment, and determines a real-time perception matrix. Based on the real-time perception matrix, it performs real-time local updates of physical elements in the farmland topology map.

[0044] Furthermore, the strategy generation component 13 is used to perform the following method: The agricultural machinery central control unit performs local obstacle detection on the updated farmland topology map based on the embedded fast judgment device and determines the judgment result; based on the judgment result, it generates obstacle avoidance instructions; based on the obstacle avoidance instructions, it activates the autonomous obstacle avoidance module embedded in the agricultural machinery central control unit, executes two-stage obstacle avoidance decisions, and generates an obstacle avoidance strategy.

[0045] Furthermore, the strategy generation component 13 is used to perform the following method: The autonomous obstacle avoidance module includes a first decision layer and a second decision layer. The first decision layer performs passable area decisions, and the second decision layer performs obstacle avoidance optimization decisions. The second decision layer has a built-in progressive obstacle avoidance mechanism and path cost function.

[0046] Furthermore, the strategy generation component 13 is used to perform the following method: Based on the updated farmland topology map, a passable area is defined; wherein, the method for defining the passable area includes: determining the crop status based on the real-time sensing matrix and assessing the agronomically permissible passable area; determining the soil conditions based on the real-time sensing matrix and assessing the actual safe passable area; determining the physical parameters of agricultural machinery based on the real-time sensing matrix and calculating the theoretically passable area; and defining the passable area by taking the intersection of the agronomically permissible passable area, the actual safe passable area, and the theoretically passable area.

[0047] Furthermore, the strategy generation component 13 is used to perform the following method: The path cost function is a weighted sum of crop damage cost, operational efficiency cost, energy consumption cost, and soil compaction cost. The crop damage cost is defined by the compaction area and damage recoverability; the operational efficiency cost is defined by the detour distance, speed loss, and impact on subsequent operations; the energy consumption cost is defined by the path curvature, ground resistance, and speed variation; and the soil compaction cost is defined based on the number of passes, soil moisture, and load weight.

[0048] Furthermore, the strategy generation component 13 is used to perform the following method: The progressive obstacle avoidance system consists of multiple obstacle avoidance modes, including path fine-tuning, speed adjustment, local replanning, and emergency stop. Obstacle avoidance boundaries are defined using the passable area. Multiple obstacle avoidance modes are determined by segmenting the path through local obstacles. For the first segment of the multiple obstacle avoidance modes, multiple rounds of iterative optimization are performed under the constraints of the path cost function and the first obstacle avoidance mode to determine the first obstacle avoidance strategy. Starting from the endpoint of the first obstacle avoidance strategy, obstacle avoidance path analysis is performed on the second segment until the Nth obstacle avoidance strategy is determined. These strategies are then combined and integrated to form the final obstacle avoidance strategy.

[0049] Furthermore, the execution management component 14 is used to execute the following methods: The obstacle avoidance strategy is decomposed into multiple sub-strategies using the smallest execution unit driven by agricultural machinery. The multiple sub-strategies are then subject to timestamp constraints and marking based on execution unit encoding to generate multi-threaded obstacle avoidance instructions. The agricultural machinery central control unit executes the agricultural machinery obstacle avoidance control by issuing the multi-threaded obstacle avoidance instructions.

[0050] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for automatic obstacle avoidance control of an agricultural power machine using multi-sensor fusion, characterized by, The method comprises: Based on the sensing array loaded by the agricultural machine, the agricultural field range detection scanning under the preset range constraint is performed under the driving of the agricultural machine, and multi-source sensing data is acquired; By uploading an agricultural task, combining a pre-defined classification mode of field obstacles and a field topological map, a rapid judgment device and an autonomous obstacle avoidance module are developed in the agricultural machine control center; By dynamically weighting and fusing the multi-source sensing data, a local obstacle detection based on the rapid judgment device is triggered, and a passable area decision and an obstacle avoidance optimization decision based on the autonomous obstacle avoidance module are generated to generate an obstacle avoidance strategy; An execution unit driven by the agricultural machine mechanically responds to the obstacle avoidance strategy to perform automatic obstacle avoidance management.

2. The multi-sensor fusion, automatic obstacle avoidance control method for an agricultural power machine of claim 1, wherein, A classification mode of agricultural prior knowledge definition is performed, wherein the classification mode classifies the field area into a hard obstacle class, a soft obstacle class, a crossable object and a work object; By agricultural task interpretation, task elements are determined, wherein the task elements include a task priority matrix and a constraint condition set; For the field area, a work object area, a mechanical travel path and a neighboring area are identified as a field topological map.

3. The multi-sensor fusion automatic obstacle avoidance control method for an agricultural power machine of claim 1, wherein, The dynamic weighting and fusion of the multi-source sensing data comprises: The multi-source sensing data is received to perform multi-source sensing dynamic weighting and fusion in the field environment to determine a real-time perception matrix; According to the real-time perception matrix, a real-time local update of physical elements in the field topological map is performed.

4. The multi-sensor fusion, automatic obstacle avoidance control method for an agricultural power machine of claim 3, wherein, The obstacle avoidance strategy is generated, comprising: The agricultural machine control center performs local obstacle detection on the updated field topological map according to the embedded rapid judgment device to determine a judgment result; According to the judgment result, an obstacle avoidance instruction is generated; According to the obstacle avoidance instruction, the autonomous obstacle avoidance module embedded in the agricultural machine control center is activated to perform two-stage obstacle avoidance decision to generate an obstacle avoidance strategy.

5. The multi-sensor fusion agricultural power machine automatic obstacle avoidance control method of claim 4, wherein, The autonomous obstacle avoidance module includes a first decision layer and a second decision layer, wherein the first decision layer performs a passable area decision, and the second decision layer performs an obstacle avoidance optimization decision, and the second decision layer is built-in with a progressive obstacle avoidance mechanism and a path cost function.

6. The multi-sensor fusion, automatic obstacle avoidance control method for an agricultural power machine of claim 5, wherein, The first decision layer performs a passable area decision, comprising: According to the updated field topological map, a passable area is defined; Wherein, the definition method of the passable area comprises: According to the real-time perception matrix, the crop state is determined to evaluate the agronomic allowable passing area; According to the real-time perception matrix, the soil condition is determined to evaluate the actual safe passing area; According to the real-time perception matrix, the physical parameters of the agricultural machine are determined to calculate the theoretical passable area; The intersection of the agronomic allowable passing area, the actual safe passing area and the theoretical passable area is taken to define the passable area.

7. The multi-sensor fusion agricultural power machine automatic obstacle avoidance control method of claim 6, wherein, The path cost function is a weighted sum based on crop damage cost, work efficiency cost, energy consumption cost and soil compaction cost; Wherein, the crop damage cost is defined by the rolling area and the damage recoverability, the work efficiency cost is defined by the detour distance, speed loss and subsequent work influence, the energy consumption cost is defined by the path curvature, ground resistance and speed change, and the soil compaction cost is defined based on the number of passes, soil moisture and load weight.

8. The multi-sensor fusion agricultural power machine automatic obstacle avoidance control method of claim 7, wherein, The progressive obstacle avoidance mechanism is composed of multiple obstacle avoidance modes, including path fine-tuning, speed adjustment, local re-planning and emergency stop. The passable area is used for obstacle avoidance boundary limitation, and a multi-section obstacle avoidance mode is determined by performing path segmentation on local obstacles; For the first segmented path in the multi-section obstacle avoidance mode, the path cost function is executed and multi-round iteration optimization is performed under the first obstacle avoidance mode constraint to determine the first obstacle avoidance strategy; The end point of the first obstacle avoidance strategy is used as the starting point to perform obstacle avoidance path analysis of the second segmented path until the Nth obstacle avoidance strategy is determined, and the obstacle avoidance strategy is integrated.

9. The multi-sensor fusion automatic obstacle avoidance control method for an agricultural power machine of claim 1, wherein, The execution unit driven by the agricultural machinery responds to the obstacle avoidance strategy to perform automatic obstacle avoidance management, including: Decompose the obstacle avoidance strategy into multiple sub-strategies with the smallest execution unit driven by the agricultural machinery; Timestamp constraint and label based on execution unit coding are performed on the multiple sub-strategies to generate multi-threaded obstacle avoidance instructions; The agricultural machinery central control executes agricultural obstacle avoidance control by issuing the multi-threaded obstacle avoidance instructions.

10. A multisensor fusion based automatic obstacle avoidance control system for agricultural power machines, characterized by, The system for implementing the automatic obstacle avoidance control method of the agricultural power machine according to any one of claims 1-9, the system comprising: Data acquisition component: based on the sensor array loaded by the agricultural machinery, the data acquisition component performs farmland range detection scanning under preset range constraint to obtain multi-source sensing data; Judgment development component: by uploading an agricultural task, combining a pre-defined classification mode of farmland obstacles and a farmland topological map, a rapid judgment device and an autonomous obstacle avoidance module are developed in the agricultural machinery central control; Strategy generation component: by dynamically weighting and fusing the multi-source sensing data, triggering local obstacle detection based on the rapid judgment device, and passable area decision and obstacle avoidance optimization decision based on the autonomous obstacle avoidance module, an obstacle avoidance strategy is generated; Execution management component: the execution unit driven by the agricultural machinery responds to the obstacle avoidance strategy to perform automatic obstacle avoidance management.

Images