Systems and methods for controlling thinning of plant propagation structures

Abstract

According to some aspects of the invention, a system for controlling thinning of plant propagation structures (e.g., fruits and / or flowers) includes a processing unit; and a non-transitory medium readable by the processing unit. The medium stores instructions that, when executed by the processing unit, cause the processing unit to generate a thinning strategy based on analysis of data, which may include one or more of: strain / cultivar, geographic area, terrain, soil, farming practice, climate, season, target market preference, and trade. The thinning strategy may include determining a desired propagation structure density, which may refer to a ratio (percentage) of flowers / fruits to be removed and a distance between flowers / fruits. Generally, the density is related to the position of fruits / flowers. The density may be a spatial density, a linear density, a tree average density, a branch average density, or an average density per unit length. The thinning strategy can be determined according to the line / cultivar and the geographical environment.

Description

Technical Field

[0001] The present invention relates to pruning, and in particular, the present invention relates to automatically controlled fruit / flower thinning. Background Art

[0002] External plant structures such as leaves, stems, roots, flowers, fruits, and seeds are called plant organs. Each organ is an organized group of tissues that work together to perform a specific function. Some of these organs are reproductive structures. These structures can be divided into two categories: sexual reproductive structures and asexual reproductive structures. Sexual reproductive structures include flower buds, fruits, and seeds.

[0003] Flower or young fruit removal (thinning) is done to allow the remaining fruit to grow quickly and to avoid such a large harvest affecting the next year's harvest. Thinning is done by manual, mechanical or chemical methods. Manual flower thinning may be more predictable than mechanical or chemical flower thinning, but it relies heavily on manpower, which can involve complications and high costs. The cost of thinning is very high compared to most operations involved in the growing process except harvesting.

[0004] Thinning has an important impact on the rate of return obtained from a plot due to its effect on the number and size of fruit. Insufficient thinning results in a decrease in average fruit size, while excessive thinning reduces the number of fruit. Therefore, yield decreases.

[0005] Market preferences may dictate that returns for a particular fruit size range are significantly higher than returns for fruit sizes outside that range. Therefore, growers need to adopt reliable high-yield technologies to meet market demands, which may allow for appropriate responses to changing environmental conditions and market trends.

[0006] Today, thinning of fruit trees such as peaches and nectarines is done manually and often following general, inaccurate and unsystematic guidelines. Sometimes, before manual thinning, there is an inaccurate mechanical thinning with a machine, which simply cuts off some flowers or fruits, but mechanical thinning is not suitable for all varieties and most farmers do not use mechanical thinning regularly. According to existing practice of combining mechanical thinning with manual thinning, a typical example is an initial mechanical thinning of a tree with 2000 flowers, which removes between 20% and 40% of the flowers, leaving about 1500 young fruits to develop (obviously, no consideration is given to the loss between the flowering and fruiting periods). If appropriate for the variety, additional mechanical thinning can also be performed to remove young fruits, in which 20% to 30% of the young fruits are removed, so that about 1100 young fruits are left on the tree. From this stage, thinning is done manually to reduce the number of fruits to a range of between 300 and 550, depending on the variety. This means that after the initial mechanical thinning described above, manual thinning at a high level of performance requires between 5 and 8 working days per worker per dunam. In Israel, the cost per worker per day is approximately 260 NIS, roughly 80 US, calculated at the exchange rate in February 2022. If no initial mechanical thinning is applied, thinning time can be up to 9 to 11 working days per dunam.

[0007] A range of problems associated with existing technologies include: the need for a large workforce that is not always available; lack of specialization of seasonal workers; distraction of human workers (e.g., by mobile phones); the thinning strategies applied are not in line with market preferences; and there is a high reliance on experience and knowledge that many farmers and workers do not possess. It is difficult to weigh all the parameters that affect the outcome of a thinning strategy. Even knowledgeable and experienced farmers may not be able to take all parameters into account. There are also difficulties in transferring knowledge from farmers to workers. Manual thinning is a slow and expensive operation. It is difficult to perform the operation at the right time, which is limited by factors such as the need to handle thinning of several varieties in parallel or other operations in the orchard where manpower is scheduled. It is crucial to perform thinning at the right time to avoid wasting tree resources on excess fruit, rather than using tree resources precisely for the right amount of fruit to allow a harvest that meets market preferences. Working time constraints.

[0008] There are cases where chemical thinning is applied, but it is not precise enough and is not suitable for all types of crops (such as peaches and nectarines). Mechanical thinning equipment is sometimes mounted on a carrier and performs random and inaccurate thinning, usually as a preliminary to manual thinning, but not usually a replacement for manual thinning. Therefore, accurate manual thinning is essential to reduce the number of fruit on the tree to the desired final number.

[0009] To overcome the above problems, a number of solutions involving the use of thinning robots have been proposed. In US20130204437, Koselka et al. proposed a robotic system for crop harvesting, pruning, culling, weeding, measuring and management. The proposed system uses machine vision, using cameras to identify and locate the fruit on each tree, the points to be pruned on the grape vines, etc. The robot can be used to measure agricultural parameters or assist in the management of agricultural resources. The camera can be coupled to an arm or other device to allow observation from inside the plant while performing the desired agricultural function. The robot moves through the field and first "maps" the location of the plant, the number and size of the fruit, and the approximate location of the fruit, or maps the bands and canes of the grape vines. Once the mapping is completed, the robot or server creates an action plan that the robot can execute. The action plan can include operations and data that specify the agricultural functions to be performed. In US10757861, Robertson et al. describe a robotic fruit picking system that includes an autonomous robot with a positioning subsystem that enables the robot to be autonomously positioned using a computer vision guidance system. The robot also includes at least one picking arm and at least one picking head, or other types of end effectors mounted on each picking arm, for cutting the stems or branches of a specific fruit or fruit bunch, or pinching the fruit or bunch. The computer vision subsystem analyzes images of the fruit to be picked or stored, and the control subsystem uses machine learning techniques to program or learn the picking strategy. The quality control (QC) subsystem monitors the quality of the fruit and grades the fruit according to size and / or quality. The robot has a storage subsystem for storing the fruit in containers for storage or transportation, or in small bags for retail. Zhang Fei et al. describe a tree cutting and sorting device in CN207083570, including a tracked walking unit, an arm, and a fiber laser. Lee Chul Hee discussed in US2016050852 a robot system with a laser beam generator to cut peduncles or thin flower clusters, and more particularly discussed an agricultural robot system for identifying fruits, flowers or buds by inputting images from a camera, and cutting peduncles or thinning flower clusters (burning or curing flowers or buds) with a laser beam, thereby achieving faster pruning, and non-contact operation to avoid damage to crops and contamination by viruses and pathogens and saving labor. Morcellet proposed a robot in FR2994057, which has: a movable structure that moves between rows of grape vines; a unit that collects images related to grape vines and branches; an image processing unit; a control unit connected to the processing unit, which guides the cutting unit to the cutting point; and a recording unit for recording images to be used to guide the laser beam to the grape vines and branches.In WO2016014552, Lyons et al. describe an end effector device and automatic selective thinning system including vision acquisition hardware, kinematics-oriented and heuristic programming, a robotic arm, and an end effector designed according to fruit science. The system is used to improve the efficiency of fruit thinning processes in orchards, such as peach thinning. By automating the mechanical process of fruit thinning, the selective fruit thinning machine can eliminate manual input and further enhance favorable flower cluster removal. Combining automation with heuristic methods provides improvements to the system. The system can be configured as a robotic arm or as a handheld system, including a battery and a switch microcontroller with a handle or wrist strap. A mechanical handheld thinning device can also be part of the system. As of February 18, 2022, Aigritec provides a robotic platform and dedicated end effector for precision thinning at https: / / www.aigritec.com / service / precision-thinning / to remove excess fruit from the tree, reducing competition between fruit, allowing sufficient growing space for the fruit and exposing the fruit to sufficient sunlight, thereby achieving better overall fruit quality. Aigritec describes the use of a robotic vision system to accurately and robustly extract geometric and semantic information from the working scene of the orchard environment - ensuring that neither the fruit nor the branches are damaged.

[0010] Despite the above-mentioned great progress in the field of agricultural robots, there are still needs that cannot be met by existing technologies. The thinning robot with a robotic arm and a fruit / flower cutter mounted on the robotic arm operates in a repetitive series of actions, which significantly reduces the operating speed. Mechanical / robotic thinning may be limited by the difficulty in accessing flowers and fruits due to the branch structure. Robot operation becomes very challenging when the branch structure is uneven. In some cases, in order to facilitate the use of the robot, the branches are pruned to limit the depth of the branch structure to allow the robot arm to reach, that is, a two-dimensional (axial) tree row design. This requires continuous efforts to maintain a branch structure suitable for robot operation. This may involve a waste of tree resources. Many crops are not suitable for the typical tree row axial design with a branch structure thickness of about 80 cm.

[0011] Looking at the existing solutions mentioned above, there does not seem to be any reference material available to provide thinning operations based on certain growing conditions and related market preferences to maximize profits.

[0012] The terms “user”, “farmer” and “grower” are used interchangeably hereinafter.

[0013] The terms "pruning unit" and "pruning system" are used interchangeably hereinafter.

[0014] The terms "geographic area" and "geographic region" are used interchangeably hereinafter.

[0015] The terms "cultivar", "variety" and "line" are used interchangeably hereinafter. Summary of the invention

[0016] Existing solutions do not seem to be able to adapt thinning to different growing conditions and related market preferences to maximize profits. The inventors have developed the present invention after years of research and experimentation to overcome the challenges associated with thinning and the shortcomings of the prior art.

[0017] According to some aspects of the present invention, a system for controlling thinning of plant reproductive structures (e.g., fruits and / or flowers) includes a processing unit; and a non-transitory medium that can be read by the processing unit. The medium stores instructions that, when executed by the processing unit, cause the processing unit to generate a thinning strategy based on an analysis of data, which may include one or more of the following: strain / cultivar, geographic region, topography, soil, farming practices, climate, season, target market preferences, and trade. The thinning strategy may include determining the desired reproductive structure density, which may refer to the ratio (percentage) of flowers / fruits to be removed and the distance between flowers / fruits. Typically, density is related to the location of fruits / flowers. Density may be spatial density, linear density, average tree density, average branch density, average density per unit length. The thinning strategy may be determined based on strain / cultivar and geographic environment. This may include parameters such as soil, climate, sunshine, latitude, altitude, drainage, soil treatment history, previous crops, fertigation methods, water quality, alluvial deposits, soil loss, sediments, groundwater, air and soil pollution, and soil conditions (such as erosion and compaction). The processing unit generates an output derived from the thinning strategy, which may be a recommendation provided to the farmer or a signal sent to control the operation of the thinning equipment.

[0018] According to some embodiments of the invention, the thinning system includes one or more reproductive structure (such as flowers or fruits) detection units coupled to the processing unit in a communicative manner. These detection units may include cameras, sensors, lidars, radars, image processing and analysis modules, machine learning modules, artificial intelligence, databases, and communication modules. The processing unit can calculate the transient reproductive structure density (including zero time) based on the input received from the reproductive structure detection unit. The thinning strategy includes inferences based on the gap between the transient density and the desired reproductive structure density.

[0019] According to some embodiments, the system includes a tree structure detection unit, whereby the processing unit generates a temporal model of one or more parts of each tree based on inputs received from the reproductive structure detection unit and the tree structure detection unit. These detection units may include cameras, sensors, lidars, radars, image processing and analysis modules, machine learning modules, artificial intelligence, databases, and communication modules.

[0020] The thinning strategy may include determining the required fruit size at harvest and / or determining the fruit size to be pruned during thinning. The strategy may include inferences based on user preferences derived from inputs received from at least one recommendation system, which may include data about one or more of the following: strains / cultivars, geographic regions, terrain, soils, irrigation, climate seasons, target market preferences, and trade. The strategy may include inferences based on inputs provided by the user, inputs received from various detectors, including cameras, thermal imaging, image processing systems, sound, radar, laser radar, etc. In addition, inputs may also be received from databases, the Internet, and weather monitoring equipment. The inputs may also include historical data. In order to process the inputs and generate the thinning strategy, the processing unit may utilize modules for machine learning, artificial intelligence, manual decision making, data mining, and processing big data.

[0021] In some embodiments of the present invention, the thinning system includes a computerized controller that receives the output from the thinning strategy and the temporal reproductive structure density from the processing unit. The system may include one or more preliminary pruning units and one or more first-level pruning units, both of which are controlled by the controller. The thinning strategy includes determining a threshold for the density of the preliminary pruning reproductive structure. If the temporal density is higher than the threshold, the preliminary pruning unit is started and operated until the temporal density is lower than the threshold, and if the temporal density is lower than the threshold, the first-level pruning unit is started and operated until the temporal density reaches the desired density value of the reproductive structure according to the thinning strategy. The first-level pruning unit may include a plurality of laser beam emitters, each of which is coupled to one or more laser beam direction and intensity control units. Some of the direction and intensity control units may be based on optical principles. In addition, since the density of the reproductive structure may be different between different parts of the tree, the pruning units may be started separately for individual parts of the tree.

[0022] The thinning system can be installed on a vehicle. To prevent possible collateral damage caused by the laser beam or other parts, the system can include protective devices, such as protective barriers; motion detectors and image analysis equipment. For example, when an object, animal or person is detected approaching the working area, or when a bird's nest or trap that should be protected is detected in the working area, detectors and imaging devices can be used to stop the operation of the thinning system. Detectors can include cameras, thermal imaging (thermal), image processing, sound, radar, lidar, and other devices that may be developed in the future. Machine learning, artificial intelligence, databases, communication modules and historical data can also be used. Some detectors can be fixed and permanently located in the working area, or they can be transferred between different working areas. Some detectors can be fixed on the thinning system, mounted on a vehicle or on an aircraft such as a drone and a balloon. According to some examples, a mobile phone connected to the thinning system is used to send a warning signal of the possibility of peripheral damage, such as an object detected by the mobile phone camera.

[0023] The pruning unit may include: one or more plant part detection units; and a processing unit for receiving input from the plant part detection unit. A non-transitory medium can be read by the pruning processing unit, and the medium stores instructions, which, when executed by the processing unit, cause the processing unit to generate one or more structural models of one or more occluding sections of the plant part based on the input from the plant part detection unit. The pruning processing unit can be used to define the optimal cutting point in the pruning of the reproductive structure based on the occluding part model and the input received from the plant part detection unit. The plant part detection unit may include cameras, thermal imaging, image processing, sound processing, radar, lidar, and other types of detectors that may be developed in the future. Machine learning, artificial intelligence, training modules, databases, communication modules, historical data, and big data can also be used.

[0024] The pruning unit may include mechanical, hydraulic, pneumatic, robotic, laser, electrical mechanisms, etc. In addition to propagation structures, some embodiments of the present invention may also be used to prune branches or other plant parts.

[0025] Methods according to some aspects of the invention involve using an artificial intelligence (AI) module, including generating a thinning strategy and generating outputs derived from the thinning strategy. The thinning strategy is based on an analysis of data, which may include strains / cultivars, geographic regions, terrain, soils, farming practices, climate, seasons, target market preferences, and trades. According to some embodiments, the method includes: determining a desired reproductive structure density; calculating a transient reproductive structure density based on input received about the reproductive structure; and generating an inference based on the difference between the transient density and the value of the desired reproductive structure density. The method may include: generating a transient model of one or more parts of each tree based on inputs about the reproductive structure and inputs about the tree structure. The method may include determining the size of the fruit at harvest and / or determining the size of the fruit to be pruned during thinning. In some examples embodying the invention, the method includes generating an inference derived from inputs received by one or more recommendation systems based on user preferences, including data about strains / cultivars, geographic regions, target market preferences, and trades.

[0026] In some cases, outputs from the thinning strategy are used to control at least one pruning system. In some embodiments of the invention, the method includes determining a threshold for preliminary pruning of reproductive structure density; performing preliminary pruning; and performing first-level pruning. Preliminary pruning is performed as long as the transient density is above or equal to the threshold. First-level pruning is performed when the transient density is below the threshold and above the desired density of reproductive structures. The first-level pruning may involve emitting laser beams from a plurality of laser beam emitters. Upon receiving at least one alarm indicating possible collateral damage due to preliminary pruning or first-level pruning, the preliminary pruning and the first-level pruning are stopped.

[0027] In some embodiments, the method includes detecting one or more plant parts and generating one or more structural models of at least one occluding segment of the one or more plant parts based on the detected plant parts. The method may include defining an optimal cutting point in pruning the reproductive structure based on the structural models of the one or more occluding segments and the detected one or more plant parts. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The preferred embodiments, features, aspects and advantages of the present invention are described herein with reference to the following drawings:

[0029] Figure 1 The fruit and fruit stalk are schematically shown partially obscured by the branches and leaves;

[0030] 2 is a schematic flow chart illustrating the operation of a thinning system having two types of pruning units according to some embodiments of the present invention;

[0031] FIG3 schematically illustrates a thinning system with a protective physical barrier and two types of pruning units according to some embodiments of the present invention;

[0032] Figure 4 A tree to be pruned is schematically shown;

[0033] Figure 5 Schematically illustrates a characterization and recommendation system according to some embodiments of the present invention;

[0034] FIG6 schematically illustrates a human-machine interface of a characterization and recommendation system according to some embodiments of the present invention.

[0035] In order to leave no doubt, the elements shown in the drawings of the present patent application are presented in a manner that enables them to be clearly understood, and the proportions, size relationships and shapes do not limit the implementation thereof in any way. DETAILED DESCRIPTION

[0036] Thinning has a significant impact on the rate of return obtained from a plot due to its effect on the number and size of fruit. Market demand may dictate that the rate of return for a particular fruit size range is significantly higher than the rate of return obtained for fruit sizes outside this range. Therefore, growers need to adopt reliable high-yield technologies to meet market demands, which may allow appropriate responses to changing environmental conditions and market trends. Today, thinning of fruit trees such as peaches and nectarines is done manually and often follows general, inaccurate and unsystematic guidelines. Sometimes, manual thinning is preceded by inaccurate mechanical thinning by machines, which simply cuts off some flowers or fruits. Problems associated with existing technologies include: the need for a large labor force that is not always available; the lack of professionalism of seasonal workers; the distraction of manual workers (e.g., by mobile phones); the mismatch between the crop and market preferences caused by the implementation of thinning strategies; and the high reliance on experience and knowledge that many farmers and workers do not have. It is difficult to weigh all the parameters that affect the outcome of thinning strategies. Even knowledgeable and experienced farmers may not be able to take all parameters into account. There are also difficulties in transferring knowledge from farmers to workers. Manual thinning is a slow and expensive operation. Difficulties in performing the operation at the right time may occur due to factors such as the need to handle thinning of several varieties in parallel or other operations in the orchard where manpower is scheduled. It may be crucial to perform thinning at the right time to avoid wasting tree resources on unnecessary excess fruit, rather than using tree resources precisely for the right amount of fruit to achieve a harvest that meets market preferences. There are cases where chemical thinning is applied, but chemical thinning is not precise and is not suitable for all types of crops (such as peaches and nectarines). Despite great advances in agricultural robots, thinning may be limited by difficulty in accessing flowers and fruits due to branch structure. In addition, robotic operation becomes very challenging in cases where branch structure is uneven. Existing solutions do not seem to be able to adapt thinning to different growing conditions and related market preferences, thereby allowing for profit maximization. The inventors have developed the present invention after years of research and experimentation to overcome the challenges associated with thinning and the shortcomings of the prior art.

[0037] According to some aspects of the present invention, a system for controlling the thinning of plant reproductive structures (such as fruits and / or flowers) includes a processing unit; and a non-transitory medium that can be read by the processing unit. The medium stores instructions that, when executed by the processing unit, cause the processing unit to generate a thinning strategy based on an analysis of inputs related to the strain / cultivar, geographic region, target market preferences, and trade data. The thinning strategy can include determining the desired reproductive structure density, which can refer to the ratio (percentage) of flowers / fruits to be removed and the distance between flowers / fruits. Typically, density is related to the location of the fruit / flower. Density can be spatial density, linear density, tree average density, branch average density, average density per unit length. The thinning strategy can be determined based on the geological and geographical environment of the strain / cultivar. This can include parameters such as soil type, climate, sunshine, latitude, altitude, drainage, soil treatment history, previous crops, irrigation methods, water quality, alluvium, soil loss, sediment, air, groundwater and soil pollution, and soil erosion and compaction. The processing unit generates an output derived from the thinning strategy. The output may be advice to the farmer or a signal to control the operation of the thinning device. A thinning system embodying the invention may include such a thinning device.

[0038] The processing unit may receive and process data related to the line / cultivar. The system may define an ideal or optimal number of fruits to be left on the tree based on historical databases, characterization and recommendation systems and observations of the tree. The system may prioritize various parameters related to thinning in order to achieve the required average fruit size according to the line / cultivar, thereby achieving an average fruit size that maximizes profits according to relevant market preferences. The priority of parameters may differ for different lines / cultivars.

[0039] Lines / cultivars may be identified through characterization questionnaires completed by farmers and / or using treatment units and plant part detection units. The number of lines / cultivars may be very large, but group cataloguing may be applied to several lines / cultivars with similar characterizations. The system may catalogued unidentified lines / cultivars using a learning module, and the characterizations of these lines / cultivars may be associated with one of the groups to allow thinning to be adjusted accordingly.

[0040] Thinning can be done with adjustments to the strategy and possible additional considerations depending on the line / cultivar, such as the position of the branch relative to the trunk; the distance of the fruiting stalk from the trunk; the position of the branch relative to the sun, such as sun exposure, shade, position relative to the sun's track, and sun angle; branch diameter and length; the position of the fruit on the branch; the quality of the fruit on the branch; young fruit content analysis, such as texture, which can be used to predict fruit maturity; predict fruit color depending on the line / cultivar; and branch robustness, which is the ability of the branch to bear the weight of the fruit. The following is an illustration of some of the possible differences in thinning strategies for different varieties related to branches of 0.9 cm diameter and 40 cm long.

[0041] In varieties that have no difficulty in fruiting, fruit may be carried every 7 cm on the same side of the branch or every 3 cm on opposite sides of the branch. Where the branch begins, it may be possible to carry more fruit, for example with two fruits facing each other at the same point. In varieties that tend to produce large fruit but do not colour well, more space will be required between fruits on the same branch to allow more fruit to be exposed to the sun. In varieties where large fruit is difficult to achieve, it may be wise to allow fewer fruits on the branch. For example, allow a distance of 9 cm between fruits (instead of 7 cm), and only on the same side of the branch, to reduce competition between fruits. In varieties that tend to produce fewer fruits, fewer fruits should be left growing on the branch. When demand is high and the expected price is high, it may be wise to leave fruit with some defects (such as appearance-related defects).

[0042] The system according to some embodiments of the present invention may include a unit for identifying fruit size and defects, which is coupled to the processing unit in a communicative manner. The unit for identifying size and defects may include cameras, thermal imaging, image processing, sound processing, radar, laser radar, and other types of detectors that may be developed in the future. Machine learning, artificial intelligence, training modules, databases, communication modules, historical data, and big data may also be utilized.

[0043] According to some embodiments, inputs about market prices and demand are obtained from characterization and recommendation systems, market analysis, competitor data, and past performance to better fine-tune thinning strategies to maximize profits. Parameters for each group of varieties with similar characteristics can be weighted based on data collected from the field, system data, and data from the characterization and recommendation system so that the correct number of fruits on the trees of each variety are produced by optimally using the parameters to generate maximum value. In some embodiments, machine learning and feedback are utilized. This can include calculating the fruit left on the tree after thinning in order to analyze and continuously improve system performance, including better utilization of input received from the recommendation system. The characterization and recommendation system can take into account user expectations and preferences, strains, trade data, geographic data, weather data, terrain data, etc. to provide valuable input for analysis by the processing unit.

[0044] In some examples, the thinning system includes a preliminary mechanical pruning unit.

[0045] The system is generating a three-dimensional model of the tree to detect young fruits and / or flowers. The system then locates areas with a large number of fruits / flowers and operates a preliminary pruning unit in these areas to partially remove the fruit load in the area, thereby allowing the first-level laser beam-based pruning unit to operate more efficiently. The processing unit evaluates the results of the preliminary pruning and its impact on the overall efficiency of the thinning. For example, if the laser unit deviates from its target efficiency according to the thinning strategy, the operation of the mechanical thinning unit will be changed to improve the target efficiency. Efficiency can be defined in a variety of ways, such as in terms of the number of fruits removed per unit time. When it is detected that the laser unit removes less fruit than the expected number of fruits, that is, its efficiency is too low, this is an example of the interaction between the performance of the primary unit and the first-level (laser) unit. In the next run, the mechanical unit will remove less fruit, leaving more fruit to be removed by the laser unit. If it is detected that the laser unit removes too much fruit, the operation of the mechanical unit will be changed to remove more fruit, leaving less fruit for the laser unit to remove.

[0046] The purpose of a laser pruning unit is generally to perform precise thinning according to a thinning strategy. In some embodiments of the present invention, the laser pruning unit is mounted on a structure, and the laser pruning unit includes a plurality of laser emitters to facilitate access to the fruit or flowers on the tree. Each laser emitter is coupled to an optical computerized laser beam guiding unit. The laser beam guiding unit receives visual data about the target size (such as the diameter of the fruit stalk), etc. The beam guiding unit can adjust the laser wavelength to accurately match the fruit stalk diameter to eliminate collateral damage, i.e., the laser beam deviates from the fruit stalk to be cut. In some examples embodying the present invention, a cutting point module unit is used to identify the target (e.g., the fruit stalk) through machine learning. The cutting point module can be coupled to an infrared thermal imager (thermal camera) for detecting the temperature difference between a tree part (e.g., a leaf) and the target to allow the laser beam to be directed to the target when the target is obscured. In some cases, a blower is used to move branches and leaves that obscure the cutting point (target). The cutting point module can utilize artificial intelligence to generate a virtual target model based on the input exposed target portion. For example, as Figure 1As shown, when only the top of the fruit branch 101 and the fruit stalk 102 and / or the fruit 103 protruding from the leaves 104 can be partially identified. The cutting point module can take into account the angle at which the fruit is suspended, the diameter of the exposed part of the fruit stalk, and statistical data to create a virtual target model. The laser unit can be coupled to a lidar to measure the distance to the target (cutting point). The camera can also be coupled to a laser system with an image processing unit to generate a three-dimensional model of the tree. The number and position of the cameras can be adjusted to allow accurate models of complex tree structures to be generated.

[0047] FIG. 2A to FIG. 2BReappearing is the operational flow chart 200 of the thinning system with two types of pruning units described above. One or more processing units receive data input by a user filling out a questionnaire 201. The data characterizes the plot to be thinned, such as line / cultivar, geographic region, topography, geology, soil, pests, vegetation index, diseases, farming practices including fertilization and irrigation, pest and disease control, crop planning, plant spacing, pollination, farming methods (such as organic farming, sustainable agriculture, service / cover crops, double cropping, precision farming), soil / crop sampling, evapotranspiration, yield, crop planting date, fertilization, tillage, vegetation cover, agronomist instructions, target market preferences, and trade. In addition, the processing unit receives input 202 from a recommendation system, which can utilize information input by the user or any information received and found relevant to characterize the relevant needs, such as information derived from the Internet of Things (IoT), telemetry, positioning (possibly GPS / GLONASS / GNSS / DGPS) aerial / satellite imaging, albedo, auxiliary data, pests and diseases, pollination, farming methods (e.g., organic farming, sustainable agriculture, service / cover crops, double-season crops, precision farming), soil / crop sampling, yield. Therefore, the information received by the processing system is not only from the user. The recommendation system can utilize big data, machine learning, metadata, fact checking and artificial intelligence. The processing unit can also receive the above data input 203 directly from the network and databases such as GIS and EROS (Earth Remote Sensing Satellite). Data 204 can be received from remote or short-range sensors, such as capacitance, pH, climate, imaging, conductivity, soil, electro-optical, geo-referencing, humidity, radar, lidar, reflectivity, electromagnetic and optical sensors. The processing unit generates a thinning strategy using the above data. This includes generating 205 a model of the tree to be thinned and determining 206 whether thinning should be started from preliminary mechanical pruning 207 based on the fruit load (density) until a load suitable for laser pruning 208 is detected. The thinning strategy includes one or more criteria, such as a desired fruit load at which to end thinning, and thinning will not continue until the desired fruit load is reached through the remaining fruit count 209. If the result of verifying 210 whether the fruit count 209 meets the thinning termination criteria is negative, the system reports 211 the results to the user and recalibrates 212 the operation of the mechanical and laser pruning units. The recalibration may include the use of machine learning and artificial intelligence. If the verification 210 whether the fruit count 209 meets the thinning termination criteria is positive, the system reports 213 the thinning results to the user and checks 214 whether this is the last tree to be thinned based on information entered by the user or obtained through other means such as databases or aerial imaging services. If this is not the last tree to be thinned, the processing of the next tree is started again, and if this is the last tree to be thinned, the thinning operation ends.

[0048] Thinning systems can include safeguards to prevent unnecessary damage from laser beams, including physical barriers made of heat-resistant materials that can absorb the energy of the laser beam. Motion detectors can be utilized, and if a foreign object approaches the danger zone, a signal from the detector is received in the system, thereby stopping the system operation. Cameras can be mounted high, stationary, or on drones to detect whether various objects are approaching the work area, so that the system operation can be stopped if a hazard is detected. Image analysis can be used to identify objects in trees, such as pest traps and bird nests, to stop the system operation. Figure 3A and Figure 3B , an example of a thinning system 300 with a protective physical barrier during laser thinning of a tree 301 is depicted in front view and top view, respectively. The system 300 includes a preliminary mechanical pruning unit 302 designed to quickly remove excess fruit before activating a main laser pruning unit 303. Both pruning units are operated by a computerized controller (not shown) according to the thinning strategy considered above, taking into account input from sensors such as cameras 304 mounted on the system 300. The laser pruning unit 303 includes a laser emitter 305 (some shown) that can be aimed at different angles to direct a laser beam 306 towards a cutting point (target) to remove excess fruit. To prevent collateral damage caused by the laser beam 306, the system 300 is equipped with a dome physical barrier 307. An image sensor 308 (such as a camera) is used for unmanned / remote navigation, while a headlight 309 can also be used when navigating in limited lighting conditions.

[0049] The thinning system may be mounted on an autonomous vehicle (such as a tractor) whose movement is controlled based on the thinning operation. The vehicle direction may be based on a global positioning system, etc. In some embodiments of the invention, the operation of the system, including landmarks of where the system is operating, is recorded. This may allow learning based on past performance associated with location, etc., sometimes in combination with input from LiDAR and motion sensors.

[0050] The present invention is not limited to pruning of reproductive structures. It can be used to prune branches. Branch pruning is usually done in winter, that is, a few months before thinning the reproductive structure. For example, Figure 4As shown, the branch pruning of the tree 301 is mainly divided into three stages: using heavy-duty power tools to quickly prune the skeleton (main) branches 402 that grow from the trunk 401 and are usually more than 60 mm in thickness; using manual and electric shears to prune the "leading" branches 403 with a thickness of 10-60 mm growing from the skeleton branches 402, and the fruit branches 101 growing from the skeleton branches 402; and using manual shears to shorten the fruit branches 101 with a diameter of 1-10 mm. Branch pruning, and especially shortening of fruit branches, can be considered as part of thinning because it reduces the number of fruits on the tree before harvesting. As with fruit / flower thinning, when shortening fruit branches, the thickness and length of the branches, as well as the fruit-bearing capacity of the dominant branches, should also be considered to prevent breakage and cracking due to overloading.

[0051] Various aspects of the invention allow the number of fruits to be adjusted according to the variety to achieve the best average fruit size, thereby maximizing profits. Existing methods are not practical to consider a large number of parameters in real time due to the limitations of human capabilities.

[0052] According to some aspects of the invention, the use of characterization and recommendation systems can allow thinning to be adjusted according to relevant markets and geographic regions, etc., in the following ways: by characterizing the user and providing him / her with customized suggestions on how to realize the economic potential of his / her orchard; relying on databases, etc., including past thinning results of the same variety, thinning results for that particular geographic region or plot, and relevant trade data including expected prices that may vary due to expected crops, local competition and imports. Fruits of preferred sizes in the market may sell for tens of percentage points higher than fruits of sizes outside the preferred range. However, it should be noted that the thinning strategy sufficient to achieve the preferred fruit size may be different for each variety. For example, on a particular variety, in order to achieve an average desired fruit size of 65 mm, 300 fruits should be left on the tree for harvesting. In another market, people may prefer smaller fruits of 55 mm to 60 mm in size. In this case, more than 380 fruits should be left for harvesting. These inputs can be fed back to the processing unit through the recommendation system, which can support decision-making specifically for each region and each farmer in real time within the solution of a single tree or branch. According to some embodiments of the invention, in order for the characterization and recommendation system to be oriented towards individual trees or branches, the trees or branches are mapped and characterized based on all the above-mentioned relevant details specific to each tree or branch, including the location of the individual tree and specific details (such as age, height, branch structure, pests and diseases). In the case of branches, the details will also include the location of the branch in the tree. In some cases, the data is embedded in the tree model. It should be understood that each user (farmer) can utilize the system for a specific variety / cultivar in different situations, such as different plantations, different seasons and different market conditions. The characterization of all trees and branches can be done through user input data and / or autonomously relying on sensor input, image processing, combined with the use of artificial intelligence and machine learning. Figure 5 As shown, according to some embodiments, in order to generate a customized thinning strategy, a processing unit 501 receives a customized input 502 from a characterization and recommendation system 503, receives an input 504 fed back by a user, and an input 505 derived from a database, a database, the Internet of Things, etc. Input 504 may include a filled-in questionnaire having data on strains, geographic environments, pests and diseases, pollination, farming methods, soil / crop sampling, yield, farming practices, target markets, and past performance. Input 505 may include data on trade, weather, seasons, terrain, the Internet of Things, telemetry, location, aerial / satellite imaging, albedo, market preferences, and crop price forecasts. FIG. 6A to FIG. 6CAn example of a human-machine interface 600, such as a computer or smartphone screen, is depicted for a user of the characterization and recommendation system in some embodiments of the present invention to fill out a questionnaire. Once the user registers with the characterization and recommendation system, he / she can fill out the details that appear in the fields to be filled out in order to receive customized input from the characterization and recommendation system. According to some examples, in some table fields, the user should select from predefined options, such as year, crop, season, quality classification, irrigation method, fertilization method, etc. In some examples, some fields are automatically filled in based on other fields or previously filled in data. For example, filling in coordinates can cause some or all fields in the farming practice table and the historical data table to be automatically filled in. FIG. 6A to FIG. 6C The questionnaire is not intended to limit the scope of the present invention, but is provided for illustration only and is not exhaustive. According to some embodiments of the present invention, the questionnaire may include additional questions or may not include additional questions. FIG. 6A to FIG. 6C The reader should understand that the present invention is not limited to user-filled forms. In some embodiments of the present invention, FIG. 6A to FIG. 6C The data shown in the examples shown can be obtained in the characterization and recommendation system or directly in the processing unit from a database or sensor. For example, the number of fruits in harvest can be obtained from an image processing system. The location of the tree can be determined based on input from one or all of the image processing system, aerial photos, satellite data, lidar, positioning system, etc.

[0053] Numerous other changes, substitutions, variations, and modifications may be ascertained by those skilled in the art, and it is intended that the present invention include all such changes, substitutions, variations, and modifications as fall within the scope of the appended claims.

[0054] In addition, those skilled in the art will recognize that the functional units in this description are labeled as modules throughout the specification. Those skilled in the art will also recognize that modules can be implemented as circuits, logic chips, or separate components of any class. In addition, those skilled in the art will also recognize that modules can be implemented in software, which is then executed by various processor architectures. In an embodiment of the present invention, a module may also include computer instructions or executable codes that may instruct a computer processor to perform a series of events based on received instructions. The choice of module implementation is left to the discretion of those skilled in the art to design and select, and does not limit the scope of the present invention in any way.

Claims

1. A system for controlling thinning of plant reproductive structures, the system comprising: Processing unit; as well as a non-transitory medium readable by the processing unit, the medium storing instructions that, when executed by the processing unit, cause the processing unit to generate a thinning strategy, wherein the strategy is based on analysis of data selected from the group consisting of: line / cultivar, geographic region, topography, soil, farming practices, climate, season, target market preferences, trade, and combinations thereof, Wherein, the processing unit generates an output obtained from the thinning strategy.

2. The system of claim 1, comprising at least one reproductive structure detection unit communicatively coupled to the processing unit, in, The thinning strategy includes determining the desired reproductive structure density, wherein the non-transitory medium is readable by the processing unit, the medium storing instructions which, when executed by the processing unit, cause the processing unit to calculate a transient reproductive structure density based on input received from the reproductive structure detection unit, and wherein the thinning strategy includes an inference based on a difference between the transient density and a value of the expected reproductive structure density.

3. The system according to claim 2, comprising a tree structure detection unit, wherein: The non-transitory medium is readable by the processing unit, the medium storing instructions that, when executed by the processing unit, cause the processing unit to generate a transient model of at least a portion of each tree based on input received from the reproductive structure detection unit and the tree structure detection unit.

4. The system according to claim 1, wherein: The strategy includes determining the desired fruit size at harvest.

5. The system according to claim 1, wherein: The strategy includes determining the fruit size to be pruned during thinning.

6. The system according to claim 1, wherein: The strategy includes inferences obtained based on user preferences according to input received from at least one characterization and recommendation system.

7. The system according to claim 6, wherein: The characterization and recommendation system for assisting the processing unit in generating a customized thinning strategy takes into account inputs received from a questionnaire distributed among farmers, wherein the questionnaire when filled in includes data selected from the following: line, geographical environment, pests and diseases, pollination, farming methods, soil / crop sampling, yield, farming practices, target market and past performance; wherein the characterization and recommendation system also utilizes data input selected from the following: trade, weather, season, terrain, Internet of Things, telemetry, location, aerial / satellite imaging, albedo, market preferences and crop price forecasts.

8. The system according to claim 1, wherein: The outputs include outputs for controlling at least one trimming system.

9. The system of claim 8, comprising the trimming system.

10. The system according to claim 2, comprising: a computerized controller that receives output from the thinning strategy and the transient reproductive structure density; at least one preliminary trimming unit, the at least one preliminary trimming unit being controlled by the controller; as well as at least one first-stage trimming unit, the at least one first-stage trimming unit being controlled by the controller, Among them, the thinning strategy includes determining a threshold value for the density of preliminary pruning reproductive structures. If the transient density is higher than the threshold, the preliminary pruning unit is started and operated until the transient density is lower than the threshold; if the transient density is lower than the threshold, the first-level pruning unit is started and operated until the transient density reaches the value of the expected reproductive structure density.

11. The system according to claim 10, wherein: The first stage trimming unit comprises a plurality of laser beam emitters, each of the emitters being coupled to at least one laser beam direction and intensity control unit.

12. The system of claim 10, mounted on a vehicle.

13. The system of claim 11, comprising at least one device for preventing possible collateral damage due to the laser beam, the device being selected from the group consisting of: a protective barrier, a motion detector, an image analysis device, and combinations thereof.

14. A pruning unit for shielding plant parts, the pruning unit comprising: at least one plant part detection unit; a processing unit that receives input from the plant part detection unit; as well as A non-transitory medium, which can be read by the processing unit, and the medium stores instructions, which when executed by the processing unit, cause the processing unit to generate at least one structural model of at least one occluding segment of a plant part based on the input from the plant part detection unit.

15. The trimming unit according to claim 14, wherein: The non-transitory medium is readable by the processing unit, the medium storing instructions which, when executed by the processing unit, cause the processing unit to define an optimal cutting point in reproductive structure pruning based on the structural model of at least one occluding segment and the input from the plant part detection unit.

16. A method for controlling thinning of plant reproductive structures using an artificial intelligence (AI) module, the method comprising: Generate thinning strategies; as well as generating the output obtained from the thinning strategy, Among other things, the strategy is based on analysis of data including line / cultivar, geographic region, target market preferences and trade.

17. The method according to claim 16, comprising: Determination of desired breeding structure density; Calculating transient breeding structure density based on the received input regarding the breeding structures; as well as An inference is generated based on a difference between the transient density and the value of the expected reproductive structure density.

18. The method according to claim 17, comprising: A transient model of at least a portion of each tree is generated based on the input regarding the reproductive structure and the input regarding the tree structure.

19. The method of claim 16, comprising determining fruit size at harvest.

20. The method of claim 16, comprising determining the size of fruit to be pruned during thinning.

21. The method of claim 16, comprising generating inferences based on user preferences from input received from at least one characterization and recommendation system.

22. The method according to claim 21, wherein: The characterization and recommendation system for assisting the processing unit in generating a customized thinning strategy takes into account inputs received from a questionnaire distributed among farmers, wherein the questionnaire when filled in includes data selected from the following: line, geographical environment, pests and diseases, pollination, farming methods, soil / crop sampling, yield, farming practices, target market and past performance; wherein the characterization and recommendation system also utilizes data inputs selected from the following: trade, weather, season, terrain, Internet of Things, telemetry, location, aerial / satellite imaging, albedo, crop price forecasts and market preferences.

23. The method according to claim 16, wherein: The outputs include outputs for controlling at least one trimming system.

24. The method according to claim 18, comprising: Determine the density threshold for initial pruning of reproductive structures; Perform initial pruning; as well as Perform the first level of pruning. wherein the preliminary pruning is performed as long as the transient density is higher than or equal to the threshold, and wherein the first level pruning is performed when the transient density is lower than the threshold and higher than the desired reproductive structure density.

25. The method according to claim 22, wherein: The first stage pruning includes cutting the reproductive structure using laser beams emitted from a plurality of laser beam emitters.

26. The method of claim 22, comprising ceasing the preliminary pruning and the first level pruning upon receiving at least one input indicating possible collateral damage caused by the preliminary pruning or the first level pruning.

27. The method of claim 16, comprising detecting at least one plant part and generating at least one structural model of at least one occluding segment of at least one plant part based on the detected plant part.

28. The method according to claim 27, comprising: An optimal cutting point in pruning the reproductive structure is defined based on the structural model of the at least one occluding segment and the detected at least one plant part.

Images