Target tracking and behavior prediction method in unmanned tractor

Abstract

The embodiment of the invention relates to the technical field of automatic driving, and discloses a method for target tracking and behavior prediction in an unmanned tractor, and the method comprises the steps: firstly, capturing a target image through a depth camera, and extracting and recognizing an obstacle and a distance thereof through semantic analysis and depth information; and in combination with the tractor driving path and the obstacle distance, calculating and correcting the driving distance length to obtain a target driving distance length. And then laser radar parameters are adjusted according to the target driving distance length and the environment temperature so as to optimize data acquisition, and feature information of the target object is acquired. Then, the feature information is used to predict an activity area of the target object, and a virtual activity area is generated. Therefore, the target identification precision and behavior prediction capability of the unmanned tractor in a complex environment are remarkably improved, and the driving safety and operation efficiency of the tractor are effectively enhanced through accurate obstacle identification and optimized data acquisition.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous driving, and particularly to a method for tracking and behavior prediction of targets in an unmanned tractor and a tractor device. Background Art

[0002] With the rapid development of agricultural intelligent technology, the application of unmanned tractors in precision agriculture is becoming increasingly widespread. Its core tasks are to achieve autonomous navigation, obstacle avoidance, and operation path optimization in complex farmland environments. However, farmland scenarios have characteristics such as variable terrain undulations (such as muddy areas and gullies), randomly appearing dynamic obstacles (such as moving agricultural machinery and livestock), and harsh environmental conditions (such as high temperature and dust), which pose extremely high requirements for the perception and decision-making capabilities of autonomous driving systems.

[0003] In related technologies, obstacle detection methods based on single vision or lidar often result in false detections or missed detections due to environmental interference (such as light changes, rain and fog occlusion), and traditional path planning algorithms mostly rely on static maps and are difficult to adapt to the real-time changes of dynamic obstacles and complex terrains. Summary of the Invention

[0004] The main objective of the present invention is to provide a method for tracking and behavior prediction of targets in an unmanned tractor and a tractor device, aiming to solve the technical problem that the target recognition method in the prior art is difficult to adapt to variable environments and results in inaccurate recognition.

[0005] To achieve the above objective, in a first aspect, an embodiment of the present application provides a method for tracking and behavior prediction of targets in an unmanned tractor, which is applied to a tractor device. The method includes: Obtain a target image by acquiring a depth image captured by a depth camera in the tractor device; Perform semantic analysis and depth information extraction on the target image to obtain a target object and corresponding distance information. The target object includes at least a first target object, and the first target object is a target obstacle in the vicinity of the driving path of the tractor device, and the vicinity driving path is spaced from the current driving path of the tractor device; Determine the driving distance length of the tractor device from the first target object according to the distance information and the driving path of the tractor device; Correct the driving distance length according to the complexity of the operation terrain to obtain a target driving distance length; Adjust the emission frequency and signal intensity of the lidar in the tractor device according to the target driving distance length and the environmental temperature to obtain a target parameter set, and obtain the feature information of the target object based on the target parameter set; Predict the activity area based on the feature information of the target object to obtain the virtual activity area of the target object.

[0006] In a possible implementation, the adjusting the emission frequency and signal intensity of the lidar in the tractor device according to the target driving distance length and environmental temperature to obtain a set of target parameters includes: Determine that the target driving distance length is less than or equal to the length threshold and the environmental temperature is less than or equal to the temperature threshold, and increase the emission frequency of the lidar and increase the signal intensity of the lidar to obtain a first set of target parameters; and / or, Determine that the target driving distance length is greater than the length threshold and the environmental temperature is higher than or equal to the temperature threshold, and decrease the emission frequency of the lidar and decrease the signal intensity of the lidar to obtain the second set of target parameters.

[0007] In a possible implementation, after predicting the activity area based on the feature information of the target object to obtain the virtual activity area of the target object, it further includes: Re-determine the target driving path of the tractor device according to the initial driving path of the tractor device and the virtual activity area.

[0008] In a possible implementation, before correcting the driving distance length according to the operation terrain complexity to obtain the target driving distance length, it further includes: Obtain the sinking depth of the tractor device tire and the humidity value of the operation soil; Input the sinking depth of the tractor device tire and the humidity value of the operation soil into the operation terrain complexity estimation mapping table to obtain the operation terrain complexity.

[0009] In a possible implementation, the correcting the driving distance length according to the operation terrain complexity to obtain the target driving distance length includes: Determine that the operation terrain complexity is greater than the complexity threshold, and perform a positive correction on the driving distance length to obtain the target driving distance length; Determine that the operation terrain complexity is less than the complexity threshold, and perform a reverse correction on the driving distance length to obtain the target driving distance length.

[0010] In a possible implementation, the tractor device includes a telescopic sensor, the telescopic sensor includes a telescopic rod and a first electrode plate and a second electrode plate provided on the telescopic rod, and the obtaining the humidity value of the operation soil of the tractor device includes: Before the tractor starts operating, control the telescopic sensor to insert into the working soil so that the first electrode plate, the second electrode plate and the soil form a target capacitor, where the soil serves as the dielectric substance of the target capacitor; Determine the current dielectric constant of the working soil according to the capacitance value of the target capacitor and the plate parameters, where the plate parameters include the distance between the plates and the area of the plates facing each other; Input the current dielectric constant of the working soil into a pre-trained humidity prediction model to obtain the humidity parameter of the working soil.

[0011] In a possible implementation manner, the inputting the current dielectric constant of the working soil into a pre-trained humidity prediction model to obtain the humidity parameter of the working soil includes: Input the current dielectric constant of the working soil into a pre-trained humidity value prediction model to obtain the target humidity value of the working soil, where the humidity value prediction model satisfies the following expression: ; In the formula, ε is the current dielectric constant of the working soil, is the reference dielectric constant of the working soil, is the dielectric constant of water, L is the target humidity value of the working soil, L 0 is the calculated value of the working soil humidity value, and Lr is the humidity correction amount.

[0012] In a possible implementation manner, the feature information at least includes the moving speed of the target object, and the predicting the virtual activity area of the target object according to the feature information of the target object includes: Obtain the current moving range of the target object, where the current moving range represents the moving range of the target object detected up to the current moment; Perform an outward expansion prediction on the current moving range according to the moving speed of the target object to obtain the virtual activity area of the target object.

[0013] In a possible implementation manner, the performing an outward expansion prediction on the current moving range according to the moving speed of the target object to obtain the virtual activity area of the target object includes: Determine the outward expansion radius according to the moving speed of the target object, where the outward expansion radius is the moving distance obtained based on the moving speed of the target object within a preset time period; Perform an equidistant outward expansion on the boundary of the current moving range with the outward expansion radius to obtain the virtual activity area of the target object.

[0014] In a second aspect, an embodiment of the present application further provides a tractor device, including: a memory and a processor, where the memory is used to store program code; the processor is used to call the program code to execute the method described in the first aspect.

[0015] Different from the prior art, a method for tracking and behavior prediction of a target in an unmanned tractor provided by an embodiment of the present application first obtains a target image through a depth camera, and then performs semantic analysis and depth information extraction on the image to identify obstacles (first target objects) in the adjacent driving path of the tractor and their distance information. Then, combining the tractor driving path and the obstacle distance, the driving distance length is calculated and corrected according to the complexity of the operation terrain to obtain the target driving distance length. Then, according to the target driving distance length and the environmental temperature, the emission frequency and signal intensity of the lidar are adjusted to optimize data collection, and the feature information of the target object is obtained based on the adjusted parameters. Finally, using this feature information to predict the activity area of the target object, the virtual activity area of the target object is obtained. In this way, the solution improves the target recognition and behavior prediction capabilities of the unmanned tractor in a complex environment, thereby enhancing the driving safety and operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention, and for those of ordinary skill in the art, other drawings can be obtained based on the structures shown in these drawings without creative efforts.

[0017] Figure 1 It is a schematic diagram of the operation scenario of the unmanned tractor in some embodiments of the present application; Figure 2 It is a schematic diagram of the operation path transformation of the unmanned tractor in some embodiments of the present application; Figure 3 It is a schematic flow chart of the method for tracking and behavior prediction of a target in an unmanned tractor in some embodiments of the present application; Figure 4 It is a schematic flow chart of step S400 of the method for tracking and behavior prediction of a target in an unmanned tractor in some embodiments of the present application; Figure 5 It is a schematic hardware structure diagram of the tractor device in some embodiments of the present application.

[0018] The implementation, functional features and advantages of the object of the present invention will be further described in conjunction with the embodiments and with reference to the drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0020] It should be noted that all directional indications (such as up, down, left, right, front, back...) in the embodiments of the present invention are only used to explain the relative position relationship and movement conditions between components in a specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications will also change accordingly.

[0021] In addition, the descriptions involving "first", "second", etc. in the present invention are only for descriptive purposes, and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of such features. In addition, "and / or" throughout the text includes three scenarios. Taking A and / or B as an example, it includes the technical solution of A, the technical solution of B, and the technical solution that both A and B are satisfied at the same time. In addition, the technical solutions between various embodiments can be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement it. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present invention.

[0022] With the rapid development of agricultural intelligent technologies, the application of driverless tractors in precision agriculture is becoming increasingly widespread. Its core tasks are to achieve autonomous navigation, obstacle avoidance, and operation path optimization in complex farmland environments. However, farmland scenarios have characteristics such as variable terrain undulations (such as muddy areas and gullies), randomly appearing dynamic obstacles (such as moving agricultural machinery and livestock), and harsh environmental conditions (such as high temperature and dust), which pose extremely high requirements for the perception and decision-making capabilities of the autonomous driving system.

[0023] In related technologies, obstacle detection methods based on single vision or lidar often result in false detections or missed detections due to environmental interferences (such as changes in lighting, rain, and fog occlusion), and traditional path planning algorithms mostly rely on static maps and are difficult to adapt to the real-time changes of dynamic obstacles and complex terrains.

[0024] As Figure 1 shown, Figure 1This is a schematic diagram of the operation scenario of an autonomous tractor in some embodiments of the present application. A lidar and a depth camera are provided on the tractor device T, and it can operate according to the planned driving path (S1 - S2 - S3 - S4 - S5). It can be understood that in the driving path of the tractor device T, there may be risks of emergency obstacles. For example, there is a target obstacle M in the current path S2 of the tractor device T; there may also be risks of non - emergency obstacles. For example, there is a target obstacle M in the adjacent path S3 or S4 of the tractor device T. In the embodiments of the present application, for the risk of emergency obstacles, the autonomous tractor's approach is to perform emergency braking or slow down; for the risk of non - emergency obstacles, the autonomous tractor's approach is to identify the behavioral characteristics of the target object in advance to predict its behavior, such as predicting the possible movement range of the target object, so as to change the operation path in advance according to the possible movement range of the target object.

[0025] As Figures 1-4 shown, the following takes the tractor device's execution of the method for tracking and behavior prediction of the target in the autonomous tractor as an example for explanation. It should be noted that although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here. Please refer to the appendix Figure 3 , the method includes the following steps S100 - step S600: S100. Obtain a target image by acquiring the depth image captured by the depth camera in the tractor device; Specifically, before acquiring the required depth image, a depth camera capable of acquiring high - quality depth images, such as a camera based on structured light or TOF (Time of Flight) technology, needs to be installed at a relatively high position on the tractor, and it is ensured that the depth camera is correctly installed and calibrated to acquire the depth image of the surrounding environment of the tractor device, so as to obtain the target image. The depth image includes image information and depth information.

[0026] S200. Perform semantic analysis and depth information extraction on the target image to obtain the target object and the corresponding distance information. The target object includes at least a first target object, and the first target object is a target obstacle in the adjacent driving path of the tractor device, and the adjacent driving path is spaced apart from the current driving path of the tractor device; Specifically, a deep learning model (such as a convolutional neural network) can be used to perform semantic segmentation on the depth image to identify different target objects (such as moving agricultural machinery, livestock, or soil mounds, etc.). Then, according to the ranging principle of the depth camera, the corresponding distance information is obtained, and the distance information from each target object to the tractor is extracted from the depth image. According to the distance information and information such as the driving path of the tractor equipment or the orientation of the target object, the target obstacles in the vicinity of the driving path of the tractor equipment are selected as the first target objects. The vicinity of the driving path refers to a driving path that is spaced apart from the current driving path of the tractor equipment. As Figure 1 shown, when the tractor equipment T is driving on the current path S2, it can not only obtain the object in the adjacent path S3 as the target object M, but also obtain the object in the adjacent path S4 as the target object M. Hereinafter, the case where the target object is in the adjacent path S3 is taken as an example to illustrate the solution, that is, the target obstacle in the path S3 is used as the first target object.

[0027] S300. Determine the driving distance length of the tractor equipment from the first target object according to the distance information and the driving path of the tractor equipment; Specifically, after obtaining the distance L1 between the tractor equipment T and the target object M, the projection distance L2 of the distance L1 in the current driving path can be first determined according to the distance L1 and the driving path of the tractor equipment. After obtaining the projection distance L2, the driving distance length of the tractor equipment from the first target object can be obtained according to the projection distance L2 and the driving path of the tractor equipment (the distance along the driving path can be obtained from the driving map). The driving distance length refers to the distance that the tractor equipment needs to travel along the driving path to reach the target object.

[0028] S400. Correct the driving distance length according to the complexity of the operation terrain to obtain the target driving distance length; It can be understood that when an unmanned tractor device is operating in farmland, it will encounter various complex terrains, such as muddy, sloping, soft soil, etc. These terrain conditions will directly affect the driving speed and braking performance of the tractor device, thus affecting the accuracy of the originally calculated driving distance length. For example, in muddy terrain, the tires may slip, resulting in the actual driving distance being longer than the theoretically calculated distance, or the braking distance increasing, requiring earlier warnings and path adjustments. Specifically, the muddy ground reduces the friction between the tires and the soil, causing the driving wheels to slip, and the actual driving distance (displacement) may be much greater than the theoretical value. For example, when the slip rate is 20%, only 0.8 meters are advanced when driving 1 meter theoretically. Soft soil weakens the braking effect, and the distance required for emergency braking may increase by 30% - 50%. When the slope increases, the tractor needs to overcome the work done by gravity, the driving speed decreases, and the time taken for the same distance is longer, possibly missing the best obstacle avoidance opportunity. Accelerating on a steep slope may lead to overspeed, and the uniform speed assumption in traditional path planning fails, requiring early adjustment of the braking distance. Based on this, due to the complexity of the terrain, it is necessary to correct the driving distance length obtained in step S300 to obtain the target driving distance length.

[0029] In one embodiment, before the step S400: correcting the driving distance length according to the complexity of the operation terrain to obtain the target driving distance length, it includes: Obtain the penetration depth of the tractor device tires and the humidity value of the operation soil; Input the penetration depth of the tractor device tires and the humidity value of the operation soil into the operation terrain complexity prediction mapping table to obtain the operation terrain complexity.

[0030] Moreover, after obtaining the operation terrain complexity, the following steps for correcting the driving distance length are executed: S410, determine that the operation terrain complexity is greater than the complexity threshold, and perform a positive correction on the driving distance length to obtain the target driving distance length; S420, determine that the operation terrain complexity is less than the complexity threshold, and perform a negative correction on the driving distance length to obtain the target driving distance length.

[0031] Specifically, the penetration depth of the tire in the operating terrain or the average penetration depth over a period of time can be obtained in real time through sensors on the tractor equipment (such as pressure sensors, displacement sensors, or vision sensors, etc.). This parameter can reflect the resistance of the terrain to the tractor's travel. And the humidity value of the current operating soil can be obtained by using a soil humidity sensor or manual detection method. Soil humidity is an important factor affecting the tractor's travel resistance and tire wear. After obtaining the penetration depth of the tractor equipment's tire and the humidity value of the operating soil, the penetration depth of the tractor equipment's tire and the humidity value of the operating soil are input into the operating terrain complexity estimation mapping table or estimation model to obtain the operating terrain complexity. In the embodiments of the present application, the mapping table is established based on a large amount of experimental data and experience, and is used to map the tire penetration depth and soil humidity value to the quantitative index of the operating terrain complexity. The mapping table can be in the form of a lookup table, a regression model, or a neural network, etc. Taking the obtained tire penetration depth and soil humidity value as inputs, the complexity score of the operating terrain is calculated through the mapping table. At the same time, a suitable complexity threshold can be set in advance according to the performance, operating requirements, and safety considerations of the tractor equipment. When the operating terrain complexity exceeds the set complexity threshold, it indicates that the terrain is relatively complex, and the tractor's travel will face greater resistance and more uncertainties. For this reason, the present application makes a positive adjustment to the driving distance length, that is, increases its estimated value, which reflects that in complex terrain, the expected distance for the tractor equipment to reach the target object may be farther than the measured distance (the time taken may be longer). On the contrary, if the operating terrain complexity is lower than the complexity threshold, it means that the terrain is relatively flat, and the tractor's travel will be smoother (the time taken may be shorter). In this case, a negative adjustment is made to the driving distance length, that is, its estimated value is reduced. When the operating terrain complexity is equal to or close to the set complexity threshold, it indicates that the terrain meets the set standard and no correction is made. In this way, such an adjustment mechanism provides more accurate data support for the tractor equipment, enabling it to more accurately adjust the parameters of the radar sensor to predict the behavior of the target object, thereby ensuring the safety and efficiency of travel.

[0032] In one embodiment, the tractor equipment includes a telescopic sensor (not shown), and the telescopic sensor includes a telescopic rod and a first electrode plate and a second electrode plate provided on the telescopic rod. Based on this, obtaining the humidity value of the operating soil of the tractor equipment includes: before the tractor starts operating, controlling the telescopic sensor to insert into the operating soil so that the first electrode plate, the second electrode plate, and the soil form a target capacitor, where the soil is the dielectric material of the target capacitor; determining the current dielectric constant of the operating soil according to the capacitance value of the target capacitor and the plate parameters, and the plate parameters include the distance between the plates and the area of the plates facing each other; inputting the current dielectric constant of the operating soil into a pre-trained humidity estimation model to obtain the humidity parameter of the operating soil.

[0033] Specifically, before the tractor starts working, the telescopic sensor is controlled to be inserted into the working soil. During the insertion process, the first electrode plate and the second electrode plate will contact the soil to form a target capacitor, in which the soil acts as the dielectric material of the capacitor. Once the target capacitor is formed, the system measures its capacitance value. The size of the capacitance value is affected by many factors, including the distance between the electrode plates, the area facing the electrode plates, and the dielectric constant of the soil. Knowing the parameters of the electrode plates (such as the distance between the plates and the area facing the plates), combined with the measured capacitance value, we can use the capacitance formula to infer the current dielectric constant of the soil. The current dielectric constant of the working soil is then input into the pre-trained humidity value estimation model to obtain the target humidity value of the working soil.

[0034] In one embodiment, the humidity value estimation model satisfies the following expression: ε=ε m *[1+(ε w -ε m ) *L 0 ],L=L 0 +L r ; where ε is the current dielectric constant of the working soil, ε m is the reference dielectric constant of the working soil, ε w is the dielectric constant of water, L is the target humidity value of the working soil, L 0 is the calculated value of soil moisture during operation, and Lr is the humidity correction value.

[0035] It should be noted that after conducting multiple experiments and analyzing relevant data, the embodiment of the present application found that the dielectric constant is positively correlated with the soil moisture value, and the inventor obtained the above humidity estimation model curve by fitting the data based on multiple experimental data. Therefore, the soil moisture value can be calculated based on the above estimation model curve. And the humidity correction amount is related to the ambient humidity. When the ambient humidity is high, the moisture in the surrounding environment will interfere with the soil humidity measurement, so that the measurement result may be too large, and the humidity correction amount is a negative value; when the ambient humidity is low, the moisture in the surrounding environment will interfere with the soil humidity measurement, so that the measurement result may be too small, and the humidity correction amount is a positive value.

[0036] S500, adjusting the emission frequency and signal strength of the laser radar in the tractor equipment according to the target driving distance length and the ambient temperature to obtain a target parameter set, and acquiring characteristic information of the target object based on the target parameter set; In one embodiment, the step of adjusting the transmission frequency and signal strength of the laser radar in the tractor equipment according to the target driving distance length and the ambient temperature to obtain the target parameter set includes: Determine that the length of the target driving distance is less than or equal to the length threshold and the ambient temperature is less than or equal to the temperature threshold, increase the emission frequency of the lidar and increase the signal intensity of the lidar to obtain a first set of target parameters; and / or, Determine that the length of the target driving distance is greater than the length threshold and the ambient temperature is higher than or equal to the temperature threshold, decrease the emission frequency of the lidar and decrease the signal intensity of the lidar to obtain the second set of target parameters.

[0037] It should be noted that the set of target parameters refers to the specific parameters of the emission frequency and signal intensity of the lidar. For example, if the emission frequency parameter is m and the signal intensity parameter is n after parameter adjustment, the set of target parameters is (m, n). And under different conditions, the set of target parameters obtained after the lidar is adjusted is different (such as the first set of target parameters and the second set of target parameters). Obtaining the characteristic information of the target object based on the set of target parameters means the characteristic information of the target object obtained by the lidar under the condition of this set of target parameters. The characteristic information can be the moving speed, acceleration or body size of the target object, etc.

[0038] Specifically, when the length of the target driving distance from the tractor equipment to the target object is small, for example, when the length of the target driving distance is less than or equal to the length threshold, it means that the time for the tractor equipment to travel to the target object is short, and it is necessary to more accurately predict the behavior of the target object to ensure the driving safety of the tractor equipment and the effectiveness of path planning. In this case, if the ambient temperature during operation is less than or equal to the temperature threshold, even if the emission frequency and signal intensity of the lidar are temporarily increased, it will not affect the performance of the lidar (the low-temperature environment has a cooling effect). Therefore, at this time, increasing the emission frequency of the lidar and increasing the signal intensity of the lidar can improve the accuracy of the tractor equipment in identifying the behavior of the target object while ensuring the service life and performance stability of the lidar.

[0039] In addition, when the length of the target driving distance from the tractor equipment to the target object is large, for example, when the length of the target driving distance is greater than the length threshold, it means that the time for the tractor equipment to travel to the target object is long. In this case, if the ambient temperature during operation is higher than or equal to the temperature threshold, even if the emission frequency and signal intensity of the lidar are temporarily decreased, it will not have a great impact on the driving safety of the tractor. Therefore, at this time, decreasing the emission frequency of the lidar and decreasing the signal intensity of the lidar can improve the service life of the lidar and ensure performance stability without affecting the driving safety of the tractor.

[0040] In other environments, such as when the length of the target driving distance is less than or equal to the length threshold, but the ambient temperature is greater than the temperature threshold, the emission frequency and signal intensity of the lidar can be not adjusted to maintain the initial performance of the lidar.

[0041] Thus, in the embodiment of the present application, the transmission frequency and signal intensity of the lidar in the tractor equipment are adjusted according to the target driving distance length and the ambient temperature to more accurately obtain the target parameter set, thereby improving the accuracy of the tractor equipment in identifying the behavior of the target object or ensuring the service life and performance stability of the lidar.

[0042] S600. Predict the activity area based on the characteristic information of the target object to obtain the virtual activity area of the target object.

[0043] In one embodiment, the characteristic information at least includes the moving speed of the target object. The step S600: Predict the activity area based on the characteristic information of the target object to obtain the virtual activity area of the target object, includes: obtaining the current movement range of the target object, where the current movement range represents the movement range of the target object detected up to the current moment; extrapolating and predicting the current movement range according to the moving speed of the target object to obtain the virtual activity area of the target object.

[0044] Specifically, first obtain the current movement range of the target object. The current movement range can be the actual movement range of the target object detected by sensors such as lidar. Then determine the extrapolation radius according to the moving speed of the target object, where the extrapolation radius is the movement distance obtained based on the moving speed of the target object within a preset time period; finally, equally expand the boundary of the current movement range by the extrapolation radius to obtain the virtual activity area of the target object.

[0045] Exemplarily, the movement trajectory of the target object can be tracked in real time by using a lidar or other sensors, and based on this, the current movement range of the target object is drawn. Then extract the moving speed of the target object from the characteristic information, which is a key parameter for predicting its future movement trajectory. Based on the moving speed of the target object, a preset time period (such as 5 seconds, 10 seconds, or 1 minute, etc.) can be set, and the distance that the target object may move within this time period, that is, the extrapolation radius, is calculated. The size of this radius directly reflects the potential activity range of the target object in the future. Finally, based on the current movement range, equally expand it according to the calculated extrapolation radius to obtain the virtual activity area of the target object. This area represents the possible activity range of the target object in the future. As Figure 2 shown, F1 is the current movement range of the target object, and F2 is the virtual activity area of the target object.

[0046] In one embodiment, after predicting the virtual activity area of the target object based on the first feature information of the target object, the method further includes: re-determining the target driving path of the tractor device according to the initial driving path of the tractor device and the virtual activity area.

[0047] Specifically, through the previous steps, the virtual activity area of the target object is obtained, which represents the possible activity range of the target object in the future for a period of time. The tractor device originally stores a preset or pre-planned initial driving path. The system needs to analyze this path to determine whether there is a potential conflict with the virtual activity area. Specifically, the virtual activity area can be compared with the initial driving path to detect whether there are overlapping or adjacent areas. If there are such areas, the tractor device may conflict with the target object when driving along the initial path. Once a potential conflict is detected, the tractor device needs to immediately re-plan the driving path. The new path should avoid the virtual activity area as much as possible to reduce the risk of collision with the target object. When re-planning the path, the tractor device can generate multiple feasible alternative paths. At this time, a path optimization algorithm needs to be used to evaluate these paths and select an optimal path as the target driving path. The optimization algorithm can consider multiple factors such as path length, driving time, energy consumption, and safety. As Figure 2 shown, one of the re-planned paths is S10 - S20 - S30 - S40 - S50.

[0048] Based on this, a method for tracking and behavior prediction of a target in an autonomous tractor provided by an embodiment of the present application first obtains a target image through a depth camera, and then performs semantic analysis and depth information extraction on the image to identify obstacles (the first target object) in the adjacent driving path of the tractor and their distance information. Then, combining the tractor driving path and the obstacle distance, the driving distance length is calculated and corrected according to the complexity of the operation terrain to obtain the target driving distance length. Then, the emission frequency and signal intensity of the lidar are adjusted according to the target driving distance length and the environmental temperature to optimize data collection, and the feature information of the target object is obtained based on the adjusted parameters. Finally, the activity area of the target object is predicted using this feature information to obtain the virtual activity area of the target object. In this way, this solution improves the target recognition and behavior prediction capabilities of the autonomous tractor in a complex environment, thereby enhancing the driving safety and operation efficiency.

[0049] As Figure 5 shown, Figure 5This is a schematic diagram of the hardware structure of a tractor device in some embodiments of the present application. The tractor device provided in the embodiments of the present application further includes a memory 1000 and a processor 2000. Among them, the memory 1000 is used to store computer-readable instructions, and the processor 2000 is used to call the computer-readable instructions to execute the method for tracking and behavior prediction of a target in the driverless tractor as described above.

[0050] Among them, the processor 2000 is used to provide computing and control capabilities to control the tractor device to execute corresponding tasks. For example, the processor 2000 controls the tractor device to execute the method for tracking and behavior prediction of a target in the driverless tractor in any of the above method embodiments. The method includes: obtaining a target image by acquiring a depth image captured by a depth camera in the tractor device; performing semantic analysis and depth information extraction on the target image to obtain a target object and corresponding distance information. The target object at least includes a first target object, and the first target object is a target obstacle located in the vicinity of the driving path of the tractor device. The vicinity driving path is spaced apart from the current driving path of the tractor device; determining the driving distance length of the tractor device from the first target object according to the distance information and the driving path of the tractor device; correcting the driving distance length according to the complexity of the operation terrain to obtain a target driving distance length; adjusting the emission frequency and signal intensity of the lidar in the tractor device according to the target driving distance length and the environmental temperature to obtain a target parameter set, and obtaining feature information of the target object based on the target parameter set; predicting an activity area based on the feature information of the target object to obtain a virtual activity area of the target object.

[0051] The processor 2000 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a hardware chip, or any combination thereof; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The above PLD can be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0052] The memory 1000, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as the program instructions / modules corresponding to the method for target tracking and behavior prediction in the driverless tractor in the embodiments of the present application. By running the non-transitory software programs, instructions, and modules stored in the memory 1000, the processor 2000 can implement the method for target tracking and behavior prediction in the driverless tractor in any of the above method embodiments.

[0053] Specifically, the memory 1000 may include a volatile memory (VM), such as a random access memory (RAM); the memory 1000 may also include a non-volatile memory (NVM), such as a read-only memory (ROM), a flash memory, a hard disk drive (HDD), or a solid-state drive (SSD), or other non-transitory solid-state storage devices; the memory 1000 may further include a combination of the above types of memories.

[0054] In summary, the tractor device of the present application adopts the technical solution of any of the above method embodiments for target tracking and behavior prediction in the driverless tractor. Therefore, it at least has the beneficial effects brought by the technical solutions of the above embodiments, which will not be elaborated here one by one.

[0055] The embodiments of the present application also provide a computer-readable storage medium, such as a memory including program code, and the above program code can be executed by a processor to complete the method for target tracking and behavior prediction in the driverless tractor in the above embodiments. For example, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CDROM), a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0056] The embodiments of the present application also provide a computer program product, which includes one or more pieces of program code, and the program code is stored in a computer-readable storage medium. The processor of the tractor device reads the program code from the computer-readable storage medium, and the processor executes the program code to complete the method steps for target tracking and behavior prediction in the driverless tractor provided in the above embodiments.

[0057] Those of ordinary skill in the art can understand that all or part of the steps to implement the above embodiments can be completed by hardware, or by hardware related to program code. The program can be stored in a computer-readable storage medium, and the storage medium mentioned above can be a read-only memory, a magnetic disk, an optical disk, etc.

[0058] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0059] Through the description of the above embodiments, those of ordinary skill in the art can clearly understand that each embodiment can be implemented by means of software plus a general hardware platform, and of course, it can also be implemented by hardware. Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above method embodiments. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0060] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention. Any equivalent structural transformation made by using the content of the specification and drawings of the present invention under the inventive concept of the present invention, or direct / indirect application in other related technical fields is included in the patent protection scope of the present invention.

Claims

1. A method for tracking and predicting the behavior of a target in an unmanned tractor, characterized in that: Applied to tractor equipment, the method comprises: Obtain a target image by acquiring a depth image taken by a depth camera in the tractor equipment; Performing semantic analysis and depth information extraction on the target image to obtain target objects and corresponding distance information, the target objects at least including a first target object, the first target object being a target obstacle in an adjacent driving path of the tractor equipment, the adjacent driving path being spaced from a current driving path of the tractor equipment; Determine the driving distance length of the tractor equipment from the first target object according to the distance information and the driving path of the tractor equipment; Correcting the driving distance length according to the complexity of the working terrain to obtain a target driving distance length; According to the target driving distance length and the ambient temperature, the emission frequency and signal strength of the laser radar in the tractor equipment are adjusted to obtain a target parameter set, and characteristic information of the target object is obtained based on the target parameter set; The activity area is predicted according to the characteristic information of the target object to obtain a virtual activity area of ​​the target object.

2. The method for tracking and predicting the behavior of a target in an unmanned tractor according to claim 1, characterized in that: The target parameter set is obtained by adjusting the emission frequency and signal strength of the laser radar in the tractor equipment according to the target driving distance length and the ambient temperature, including: Determining that the target driving distance length is less than or equal to a length threshold and the ambient temperature is less than or equal to a temperature threshold, increasing the transmission frequency of the laser radar and increasing the signal strength of the laser radar to obtain a first target parameter set; and / or, Determine that the target driving distance length is greater than a length threshold and the ambient temperature is higher than or equal to a temperature threshold, reduce the transmission frequency of the laser radar and reduce the signal strength of the laser radar to obtain the second target parameter set.

3. The method for tracking and predicting the behavior of a target in an unmanned tractor according to claim 1, characterized in that: After the activity area prediction is performed according to the feature information of the target object to obtain the virtual activity area of ​​the target object, the method further includes: The target driving path of the tractor equipment is re-determined according to the initial driving path of the tractor equipment and the virtual activity area.

4. The method for tracking and predicting the behavior of a target in an unmanned tractor according to claim 1, characterized in that: Before the driving distance length is corrected according to the complexity of the working terrain to obtain the target driving distance length, the method further includes: Obtain the immersion depth of the tractor equipment tires and the moisture value of the working soil; The depth of tractor equipment tire immersion and the moisture value of the working soil are input into the working terrain complexity estimation mapping table to obtain the working terrain complexity.

5. The method for tracking and predicting the behavior of a target in an unmanned tractor according to claim 4, characterized in that: The step of correcting the driving distance length according to the complexity of the working terrain to obtain a target driving distance length includes: Determining that the complexity of the working terrain is greater than a complexity threshold, and performing a positive correction on the driving distance length to obtain a target driving distance length; It is determined that the complexity of the working terrain is less than a complexity threshold, and a reverse correction is performed on the driving distance length to obtain a target driving distance length.

6. The method for tracking and predicting the behavior of a target in an unmanned tractor according to claim 4, characterized in that: The tractor equipment includes a telescopic sensor, the telescopic sensor includes a telescopic rod and a first electrode plate and a second electrode plate provided on the telescopic rod, and the method of obtaining the moisture value of the soil in which the tractor equipment is working includes: Before the tractor starts working, controlling the telescopic sensor to be inserted into the working soil so that the first electrode plate, the second electrode plate and the soil form a target capacitor, wherein the soil serves as a dielectric material of the target capacitor; Determining the current dielectric constant of the working soil according to the capacitance value of the target capacitor and the plate parameters, wherein the plate parameters include the distance between the plates and the area facing the plates; The current dielectric constant of the working soil is input into a pre-trained humidity estimation model to obtain humidity parameters of the working soil.

7. The method for tracking and predicting the behavior of a target in an unmanned tractor according to claim 4, characterized in that: The step of inputting the current dielectric constant of the working soil into a pre-trained humidity estimation model to obtain the humidity parameter of the working soil includes: The current dielectric constant of the working soil is input into a pre-trained humidity value estimation model to obtain a target humidity value of the working soil, wherein the humidity value estimation model satisfies the following expression: ； Where ε is the current dielectric constant of the working soil, is the reference dielectric constant of the working soil, is the dielectric constant of water, L is the target humidity value of the working soil, L0 is the calculated value of the working soil humidity value, and Lr is the humidity correction value.

8. The method for tracking and predicting the behavior of a target in an unmanned tractor according to claim 1, characterized in that: The characteristic information at least includes the moving speed of the target object, and the step of performing activity area prediction according to the characteristic information of the target object to obtain a virtual activity area of ​​the target object includes: Acquire a current motion range of the target object, where the current motion range represents a motion range of the target object detected up to a current moment; The current motion range is predicted to be expanded outward according to the motion speed of the target object to obtain a virtual activity area of ​​the target object.

9. The method for tracking and predicting the behavior of a target in an unmanned tractor according to claim 8, characterized in that: The step of performing an outward expansion prediction on the current motion range according to the motion speed of the target object to obtain a virtual activity area of ​​the target object includes: Determine an outward expansion radius according to the moving speed of the target object, wherein the outward expansion radius is a moving distance obtained based on the moving speed of the target object within a preset time period; The boundary of the current motion range is expanded outward at an equal distance with the outward expansion radius to obtain a virtual activity area of ​​the target object.

10. A tractor device, characterized in that: include: A memory and a processor, wherein the memory is used to store program codes; The processor is used to call the program code to execute the method according to any one of claims 1 to 9.

Images