AUTONOMOUS NAVIGATION METHOD AND MODULE FOR AUTONOMOUS OBJECTS IN A CONTROLLED AGRICULTURAL ENVIRONMENT

Abstract

Autonomous navigation method for autonomous objects in a controlled agricultural environment, comprising: obtaining a plant map, said map containing locations of predetermined plants Within the controlled agricultural environment; obtaining plant data, said plant data associating a respective unique combination of plant features to each predetermined plant; identifying features of one or more plants in the surrounding environment of the autonomous object; identifying one or more specific plants among the predetermined plants based on the identified plant features and the plant data; and retrieving from the plant map the location of the autonomous object based on the one or more identified plants. Figure 3

Description

FIELD The present disclosure relates to an , and a training method for training the autonomous navigation module. BACKGROUND Over the years, significant infrastructure has been developed for controlled environment agriculture (CEA) to optimize plant growth, including pipe rails, racks, mounts, fixtures, etc., to support plant growth. Such infrastructure elements areknown to cause issues such as interference or disturbance ofcommunication signals. Accordingly, wireless communication in suchCEA environments is challenging. In addition,CEA environments are ofrepetitive nature, comprising (near-) identical rows and columns of racks, to optimize growth vis-a-vis oor space. Considering these factors, it is difficult for an autonomous object, such as an autonomous vehicle, to determine its position within theCEA environment based on their environment. Presently, autonomous objects rely on additional infrastructure to determine their position, such as electrical beacons (e.g., induction wires), QR-codes, wireless chargers and signals used for localization (e.g., triangulation), or manual position correction. Localization ofautonomous objects withing theCEA environment is ofhigh importance for the execution of various tasks by these objects in various locations within the environment (e.g., deleafing (removal of leaves), growth supervision, harvesting, packaging the harvest, pest and disease detection, etc. Localization is also important for data collection, as data need to be assigned to a location or object within the environment, e.g., for mapping and characterizing the environment. Furthermore, it allows for route optimization and activity planning. Several drawback of current solutions are, inter alia, installation costs, maintenance costs, accuracy, complexity, and positioning errors based, for instance, on the aforementioned disturbances in wireless communication. The repetitiveness of theCEA environment poses a further challenge in localization of an autonomous object. The state ofthe art proposed solutions such as markers placed on oors, ceilings, and / or racks or trolleys. However, oors become littered with (biological) materials (e.g., leaves, dirt), and the ceiling and racks / trolleys are not always visible due to plant growth and / or lighting conditions. Accordingly, there is a need for improvement in the localization and navigation ofautonomous objects inCEA environments. SUMMARY In view ofthe above, in a first aspect, the present disclosure provides an autonomous navigation (or localization) method for autonomous objects, such as autonomous vehicles, in a controlled agricultural environment, said method comprising: obtaining a plant map, said map containing locations ofpredetermined plants within the controlled agricultural environment, obtaining plant data, said plant data associating a respective unique combination ofplant features to each predetermined plant, identifying features ofone or more plants in the surrounding environment of the autonomous object, identifying one or more specific plants among the predetermined plants based on the identified plant features and the plant data, retrieving from the plantmap the location of the autonomous object based on the one or more identified plants. The method may be a computer-implemented method. The method may also be implemented in a non-autonomous or semi-autonomous object, as a supplementary measure. Even between plants of the same species, subtle differences occur due to environmental conditions and events, or gene variations or the like. Accordingly, each individual plant is unique in their characteristics. For instance, in number of leaves, height, width, number of stems, color, spots, orientation, temperature, soil humidity, et cetera. By mapping each plant and recording their features, it is possible to determine a position within the plantmap by recognizing these features and matching them with the plant data. Accordingly, it is possible for an autonomous object to localize itself within the controlled agricultural environment, despite its apparent repetitiveness. The method may further comprise the steps of: obtaining environment data, associating object features to predetermined inanimate objects in the controlled agricultural environment, such as mounting structures, soil blocks, pipes, wires, beams, oor signs, oor structures, building structures, signs, and marks; obtaining an environmentmap containing locations ofthe predetermined inanimate objects in the controlled agricultural environment; identifying object features related to inanimate objects in the surrounding environment of the autonomous object; identifying one or more specific inanimate objects among the predetermined inanimate objects based on the identified object features and the environment data; retrieving from the environmentmap the location of the autonomous object based on the one or more identified specific inanimate objects. The environmentmap and the plantmap may be combined, such that the plantmap also comprises the environment map. The plant data and environment data may be combined, such that the plant data also comprises the environment data. Accordingly, the step ofretrieving from the plantmap the location of the autonomous object based on the one or more identified plants, may also include concurrently retrieving from the environmentmap the location of the autonomous object based on the one or more identified specific inanimate objects, as the environmentmap may be part of the plant map. Identifying features of plants in the surrounding environment of the autonomous objectmay comprise obtaining environmental data from one or more sensors and generating identified features among a list ofpredetermined features based on the obtained environmental data. This allows the autonomous object to identify at least part of the features of a plant depending, for instance, on the range and field ofview of the sensor. Even a partial characterization of a plantmay provide sufficient information for identifying a particular plant. In the case wherein there is no unique solution, since multiple plants match the identified features, the location of the autonomous object may still be determined by observing another plant in the vicinity. Combining the partial characterization ofthe one plant and the partial characterization of the another plant with the plant map may be sufficient to converge on a location on the plant map. In a simple example, one plant may comprise twelve leaves and another plant, on the right of the one plant, fifteen leaves. Even though multiple plants in the environmentmay comprise twelve leaves, and multiple plants may comprise fifteen leaves, the plant map, combined with these identified features, may conclude that there is only one plant with twelve leaves which has another plant on its right with fifteen leaves, thereby successfully localizing the autonomous object despite a partial non-unique characterization of the plants in the surrounding environment of the object. In addition, odometry may be utilized to converge on a location on the plant map. Furthermore, the method may comprise using Kalman filtering, wherein localizations based on the identification ofone or more specific plants, the identification ofone or more specific inanimate objects, and / or odometry may be combined to obtain the best estimate of the location of the autonomous object. For instance, an extended or unscented Kalman filtermay be utilized. The environmental datamay comprise any one or more of the following: a camera image, a radar image, aLIDAR image, an ultrasound image, a humidity value , a biofeedback value, a temperature. Data sources may additionally or alternatively include a multispectral image sensor in the visible monochrome, color, near-infrared / NIR, far infrared,UV spectrum, a depth image, time- of ight sensor, and odometry data. Identifying features may comprise selecting one or more classes, and optionally subclasses, from a plurality ofpredetermined classes, optionally predetermined subclasses, for classifying features of plants, preferably based on environmental data from one or more sensors. For instance, a first class may be related to the foliage features of the plant, and a second class to the growth features of the plant. The first class may have a first subclass relating to the leaves of the plant, and a second subclass relating to the coloring ofthe plant. The second class may have a first subclass relating to the size ofthe plant, and a second subclass relating to the yield of the plant (e.g., number of owers, fruits, or vegetables). The method may further comprise, after identification of a specific plant, updating the plant data of said plant based on the identified features of said plant. As a plant constantly changes over time, tracking these changes ensures that the plant map remains correlated with the plant data, such that navigation through the method remains possible in a changing environment. The method may further comprise updating the plant data based on one or more models predicting feature evolution over time. For instance, itmay be known that a certain plant has a predictable growth in height over time, which may then be taken into account (e.g., 1 cm of heigh per day). Plant features may comprise any one or more of the following: foliage features including leaves size, leaves patterns, leaves tips, cuts and bruises on the plants, marks, spots, color differences, color contrast; growth features including size, number ofowers, fruits and vegetables, growth direction, number of leaves, leaf shape, spread, growing tip, stem features; features related to inanimate objects in the environment, including mounting structures, soil blocks, pipes, wires, beams, oor signs, oor structures, building structures, signs, and marks; biofeedback features including sap ow, soil humidity, pest and disease features; local environmental features including air temperature, air humidity, smell or sound. Sounds can be in the audible or the non-audible spectrum. Foliage features may include multispectral or non-visible information. The method may further comprise, after retrieving the location of the autonomous object based on the one or more identified plants, updating the plantmap based on the location of the autonomous object and the one or more identified plants. For instance, one plantmay have been moved with respect to another plant, and identifying the one and the another plant reveals that the plantmap need to be updated accordingly. Updating the plantmap may further be based on one or more previously retrieved locations of the autonomous object and respective one or more previously identified plants. Updating the plantmap may involve Simultaneous Localization And Mapping (SLAM) techniques. For example, due to physical activity in the controlled agricultural environment, plants may have shifted or otherwise moved as compared to a current version of the plant map. The autonomous objectmay record its position (i.e., location) with respect to a moved plant (i.e., moved as compared to the location on the plantmap of the current version), thereby introducing a positioning error. This error may become apparent at a later moment in time, when the location of the autonomous object is updated in accordance with further one or more plants. Accordingly, it can be deduced, e.g., through loop closure or the like, which plant was moved, and by how much. Hence, the plantmap can be updated accordingly. In a second aspect, the disclosure provides an autonomous navigation module for autonomous objects in a controlled agricultural environment, said module comprising: amemory configured for storing a map of the environment, said map containing locations ofpredetermined plants, and for storing plant data, said plant data associating a respective unique combination ofplant features to each predetermined plant, and a processing means configured for: identifying features of a plant in the surrounding environment of the autonomous object, identifying a specific plant among the predetermined plants based on the identified plant features and the plant data, retrieving from the map the location of the autonomous object based on the identified plant. The location may be stored in the memory, for instance in chronological and / or topological order. The module may further comprise one or more sensors for obtaining environmental data, said processing means being further configured for generating identified features among a list of predetermined features based on the obtained environmental data. The one or more sensors may comprise any one of the following: an optical sensor such as a photodetector or an image sensor, a radar, aLIDAR, an ultrasonic sensor, a humidity sensor, a biofeedback sensor, a temperature sensor. Data sources may additionally or alternatively include a multispectral image sensor in the visible monochrome, color, near-infrared / NIR, far infrared,UV spectrum, a depth image, time-of ight sensor, and odometry data. The processing means may further be configured for generating identified features including processing and / or fusing obtained environmental data from one or more data sources and / or using a feature model based on the obtained environmental data. Memory may be configured for storing a plurality ofpredetermined classes, optionally predetermined subclasses, for classifying features of plants, the processing means being further configured for identifying features by selecting one or more classes, and optionally subclasses, from the plurality ofpredetermined classes, optionally predetermined subclasses, preferably based on environmental data from one or more sensors. The processing means may further be configured for, after identification of a specific plant, updating the plant data of said plant based on the identified features of said plant. As a plant constantly changes over time, tracking these changes ensures that the plantmap remains correlated with the plant data, such that navigation through the method remains possible in a changing environment. The processing means may further be configured for updating the plant data based on one ormode models predicting feature evolution over time. Plant features comprise any one or more of the following: foliage features including leaves size, leaves patterns, leaves tips, cuts and bruises on the plants, marks, spots, color differences, color contrast; growth features including size, number ofowers, fruits and vegetables, growth direction, number of leaves, leaf shape, spread, growing tip, stem features; features related to inanimate objects in the environment, including mounting structures, soil blocks, pipes, wires, beams, oor signs, oor structures, building structures, signs, and marks; biofeedback features including sap ow, soil humidity, pest and disease features; local environmental features including air temperature, air humidity, smell or sound. Sounds can be in the audible or the non-audible spectrum. Foliage features may include multispectral or non-visible information. The processing means may further be configured for, after retrieving the location of the autonomous object based on the one or more identified plants, updating the plantmap based on the location ofthe autonomous object and the one or more identified plants. Updating the plantmap may further be based on one or more previously retrieved locations of the autonomous object and respective one or more previously identified plants. Updating the plantmap may involve Simultaneous Localization And Mapping (SLAM) techniques. For example, due to physical activity in the controlled agricultural environment, plants may have shifted or otherwise moved as compared to a current version of the plant map. The autonomous objectmay record its position (i.e., location) with respect to a moved plant (i.e., moved as compared to the location on the plantmap of the current version), thereby introducing a positioning error. This error may become apparent at a later moment in time, when the location of the autonomous object is updated in accordance with further one or more plants. Accordingly, it can be deduced, e.g., through loop closure or the like, which plant was moved, and by how much. Hence, the plantmap can be updated accordingly. In a third aspect, a training method for training the autonomous navigation module of the second aspect is provided, the training method comprising: obtaining a training dataset ofplant data associating a respective unique combination of environmental data to a predetermined feature; training a feature model based on said training data set, to generate identified features based on obtained environmental data. BRIEFDESCRIPTIONOFTHEDRAWINGS Several aspects of the present disclosure are further elucidated through the following figures, wherein: FIG. 1 shows a schematic top view of a controlled environment agriculture (CEA) environment with autonomous objects navigating therein; FIG. 2 shows a system overview of an autonomous object with an autonomous navigation module; and FIG. 3 shows aow diagram ofan autonomous navigation method. DETAILED DESCRIPTION In FIG. 1, a schematic top view of a controlled environment agriculture (CEA) environment 70 is shown, which may for instance be a greenhouse. The environment 70 may comprise rows and columns ofplant beds 50. Autonomous objects 200, such as autonomous vehicles 200 in the present example, may be utilized in the environment 70. The vehicles 200 may typically navigate in between the rows 50 and on alleys 60 communicating between rows 50. The environmentmay further comprise utilities for the vehicles 200, such as charging stations 91, a central controller 92 configured to communicate wirelessly with the vehicles 200 (e.g., for exchanging data between the vehicle 200 and a central database connected with the controller 92), one or more local beacons 93 located in the environment atknown locations, comprising communication means to communicate with the one or more autonomous vehicles 200. The location ofthe one or more beacons 93 may also be known in advance and preferably stored in a dedicated database. A charging station 91 may typically be located in between rows ofplants 50 in a greenhouse 70. Alternatively a charging station 91 may be located in an alley 60 communicating between rows of plants 50. Preferably the charging station 91 may be located at a location the vehicle 200 would navigate during its normal operation. In this way, the charging process may be performed during normal operation, avoiding unnecessary travels to a dedicated charging location. The vehicles 200 further comprise an autonomous navigation module 100 that is further described below in relation to FIG. 2. It is noted that FIG. 1 schematically depicts rows of plants 50 with similar appearance (e.g., of a similar variety or species). In practice, several different species or varieties ofplants may be present in one environment 70, and potentially within arow 50. FIG. 2 shows a system overview of an autonomous object 200, such as an autonomous vehicle 200 as described in relation to FIG. 1, with an autonomous navigation module 100 (which may also be referred to as an autonomous localization module 100). The autonomous navigation module 100 can be used for autonomous navigation (or localization) for autonomous objects 200 in a controlled agricultural environment 70. The module 100 comprises a memory 10 and a processor 20 (i.e., a processing means). The memory 10 is configured for storing a plantmap 11 of the environment 70, said map 11 containing locations ofpredetermined plants 50, and for storing plant data 12, said plant data 12 associating a respective unique combination ofplant features to each predetermined plant 50. The processor 20 is in electronic communication with the memory 10. The autonomous object 200 is equipped with one or more sensors 30 (e.g., sensor 1, sensor 2, ..., sensor n), that are communicatively coupled with the processor 20 of the autonomous navigation module 100 through communication lines 35-1, 35-2, ..., 35-n. The sensors 30 are configured for collecting data from the environment of the autonomous object 200. Particularly, the sensors 30 are configured for identifying features ofone or more plants 50 in the surrounding environment 70 of the autonomous object 200. Particularly, the processor 20 processes the sensor data to identify said features of said one or more plants 50. The processor 20 is further configured to identify a specific plant 50 among the predetermined plants 50 based on the identified plant features and the plant data 12 stored in the memory 10. The processor 20 is further configured to retrieve from the plant map 11 the location ofthe autonomous object 200, based on the identified / specific plant. For instance, through various sensor data from the sensors 30 and the plant data 12, the processor 20 may identify a plurality of specific plants 50, thereby making it possible to derive a precise position ofthe autonomous object 200. The memory 10 may further comprise feature model(s) 13, which is used to generate identified features of a plant amount a list ofpredetermines features based on obtained environmental data (i.e., sensor data from the sensors 30). The autonomous object 200may further comprise a separate navigation system 40 thatmay utilize the retrieved location of the autonomous object 200 to plan a route through the environment 70. Additionally, the navigation system may be able to navigate by means of further signals present in the environment 70. In FIG. 3 aow diagram of an autonomous navigation method 300 is shown. The method 300 may comprise five steps S1 S5, that may be partially or fully consecutive, partially or fully concurrent, and / or executed in any order. The method 300 may be a computer-implemented method. The method 300 may be used in the autonomous navigation module 100 ofFIGS. 1 and 2. In a first step S1, a plantmap 11 of the environment 70 is obtained. The map 11 may contain locations ofpredetermined plants 50 within a controlled agricultural environment 70. In a second step S2, plant data 12 is obtained. The plant data 12 may be associating a respective unique combination ofplant features to each predetermined plant 50. In a third step S3, features ofone or more plants 50 in the surrounding environment 70 of the autonomous object 200 are identified. The third step S3 may include obtaining environmental data from one or more sensors 30 and generating identified features among a list ofpredetermined features based on the obtained environmental data. Generating identified features may further comprise processing and / or fusing obtained environmental data from different data sources and / or using a feature model 13 based on the obtained environmental data. The third step S3 may include selecting one or more classes, and optionally subclasses, from a plurality ofpredetermined classes, optionally predetermined subclasses, for classifying features ofplants 50, preferably based on environmental data from one or more sensors 30. In a fourth step S4, one or more specific plants 50 are identified among the predetermined plants 50 based on the identified plant features and the plant data 12. After completing the fourth step S4, the plant data 12 of said specific plant 50 may be updated based on the identified features of said plant 50. In a fifth step S5, the location of the autonomous object 200 is retrieved from the plant map 11 based on the one or more identified plants 50. After completing the fifth step S5, the plantmap 11 may be updated based on the location of the autonomous object 200 and the one or more identified plants 50. Updating the plant map 11 may further be based on one or more previously retrieved locations of the autonomous object 200 and respective one or more previously identified plants 50. In a simple example, the method 300 is employed in an environment 70 comprising various rows and columns of plants 50. An exemplary two-dimensional plantmap 11 may be as follows: Row i / Column _» A B C 1 Plant 1A Plant 1B Plant 1C 2 Plant2A Plant 2B Plant 2C 3 Plant 3A Plant 3B Plant 3C The accompanying plant data 12may in this example be as follows: __ Suppose that in the present simple example, the autonomous object 200 is an autonomous vehicle 200 with a left-facing sensor 30, being a camera. The camera registers two plants 50. The first plant, on the right-hand side of the camera image, comprises several features including four leaves, four owers, and a height of 5 cm. The second plant, on the left-hand side of the camera image, comprises several features including three leaves, two owers, and a height of 15 cm. The processor 20 ofthe autonomous navigation module 100 of the vehicle 200 identifies these features, and finds plants with corresponding features in the plant data 12. In the present example, the first plant matches with Plant 2B, and the second plant matches with Plant 2C. These can be looked up in the plant map 11. The orientation of the camera of the vehicle 200 is known, such that the vehicle 200 (particularly the processor 20) may now conclude that it is located between Row 1 andRow 2, and between Column B and Column C, facing in the direction towards the ColumnA side ofthe map 11. Accordingly, the autonomous navigation module 100 has successfully localized the vehicle 200 in the environment 70. Whilst the principles ofthe invention have been set out above in connection with specific embodiments, it is understood that this description is merely made by way ofexample and not as a limitation of the scope ofprotection which is determined by the appended claims. CLAUSES The disclosure comprises the following clauses that correspond exactly to the Dutch-language claims: 1. An autonomous navigation method for autonomous objects in a controlled agricultural environment, said method comprising: obtaining a plant map, said map containing locations ofpredetermined plants within the controlled agricultural environment, obtaining plant data, said plant data associating a respective unique combination ofplant features to each predetermined plant, identifying features ofone or more plants in the surrounding environment of the autonomous object, identifying one or more specific plants among the predetermined plants based on the identified plant features and the plant data, retrieving from the plantmap the location of the autonomous object based on the one or more identified plants. 2. The autonomous navigation method of clause 1, wherein identifying features of plants in the surrounding environment of the autonomous object comprises obtaining environmental data from one or more sensors and generating identified features among a list of predetermined features based on the obtained environmental data. 3. The autonomous navigation method of clause 2, wherein the environmental data comprises any one or more of the following: a camera image, a radar image, aLIDAR image, an ultrasound image, a humidity value , a biofeedback value, a temperature. 4. The autonomous navigation method of any of the above clauses, wherein identifying features comprises selecting one or more classes , and optionally subclasses, from a plurality ofpredetermined classes, optionally predetermined subclasses, for classifying features of plants, preferably based on environmental data from one or more sensors. 5. The autonomous navigation method of any of the above clauses, further comprising after identification of a specific plant, updating the plant data of said plant based on the identified features of said plant. 6. The autonomous navigation method of any of the above clauses, further comprising updating the plant data based on one or more models predicting feature evolution over time. 7. The autonomous navigation method of any of the above clauses, wherein plant features comprise any one or more of the following: foliage features including leaves size, leaves patterns, leaves tips, cuts and bruises on the plants, marks, spots, color differences, color contrast; growth features including size, number ofowers, fruits and vegetables, growth direction, number of leaves, leaf shape, spread, growing tip, stem features; features related to inanimate objects in the environment, including mounting structures, soil blocks, pipes, wires, beams, oor signs, oor structures, building structures, signs, and marks; biofeedback features including sap ow, soil humidity, pest and disease features; local environmental features including air temperature, air humidity, smell or sound. 8. The autonomous navigation method of any of the above clauses, further comprising: after retrieving the location of the autonomous object based on the one or more identified plants, updating the plantmap based on the location of the autonomous object and the one or more identified plants. 9. The autonomous navigation method of clause 8, wherein updating the plantmap is further based on one or more previously retrieved locations of the autonomous object and respective one or more previously identified plants. 10. An autonomous navigation module for autonomous objects in a controlled agricultural environment, said module comprising: amemory configured for storing a map of the environment, saidmap containing locations ofpredetermined plants, and for storing plant data, said plant data associating a respective unique combination ofplant features to each predetermined plant, a processing means configured for: o identifying features of a plant in the surrounding environment of the autonomous object, o identifying a specific plant among the predetermined plants based on the identified plant features and the plant data, o retrieving from the map the location of the autonomous object based on the identified plant. 11. The autonomous navigation module of the previous clause, further comprising one or more sensors for obtaining environmental data, said processing means being further configured for generating identified features among a list ofpredetermined features based on the obtained environmental data. 12. The autonomous navigation module of the previous clause, wherein the one or more sensors comprise any one of the following: an optical sensor such as a photodetector or an image sensor, a radar, aLIDAR, an ultrasonic sensor, a humidity sensor, a biofeedback sensor, a temperature sensor. 13. The autonomous navigation module of clause 9 or 10, the processing means being further configured for generating identified features further comprises processing and / or fusing obtained environmental data from one or more data sources and / or using a feature model based on the obtained environmental data. 14. The autonomous navigation module of any of the above clauses, wherein memory is configured for storing a plurality ofpredetermined classes , optionally predetermined subclasses, for classifying features of plants, the processing means being further configured for identifying features by selecting one or more classes, and optionally subclasses, from the plurality ofpredetermined classes, optionally predetermined subclasses, preferably based on environmental data from one or more sensors. 15. The autonomous navigation module of any of the above clauses, the processing means being further configured for, after identification of a specific plant, updating the plant data of said plant based on the identified features of said plant. 16. The autonomous navigation module of any of the above clauses, the processing means being further configured for updating the plant data based on one or mode models predicting feature evolution over time. 17. The autonomous navigation module of any of the above clauses, wherein plant features comprise any one or more of the following: foliage features including leaves size, leaves patterns, leaves tips, cuts and bruises on the plants, marks, spots, color differences, color contrast; growth features including size, number ofowers, fruits and vegetables, growth direction, number of leaves, leaf shape, spread, growing tip, stem features; features related to inanimate objects in the environment, including mounting structures, soil blocks, pipes, wires, beams, oor signs, oor structures, building structures, signs, and marks; biofeedback features including sap ow, soil humidity, pest and disease features; local environmental features including air temperature, air humidity, smell or sound. 18. A training method for training the autonomous navigation module according to any of the above module clauses, comprising: obtaining a training dataset ofplant data associating a respective unique combination of environmental data to a predetermined feature; training a feature model based on said training data set, to generate identified features based on obtained environmental data.

Claims

1. Autonomous navigation method for autonomous objects in a controlled agricultural environment, comprising the working method: - obtaining a plant map, where the map shows the locations of predetermined plants within the controlled agricultural environment contains, - obtaining plant data, where the plant data has a respective unique associate combination of plant characteristics with each predetermined plant, - identifying characteristics of one or more plants in the surrounding environment of the autonomous object, - identifying one or more specific plants among the predetermined plants based of the identified plant characteristics and the plant data, - retrieving the location of the autonomous object from the plant map based on the one or more identified plants.

2. Autonomous navigation method within the meaning of claim 1, involving the identification of features of plants in the surrounding environment of the autonomous object obtaining includes environmental data from one or more sensors, as well as the generation of identified characteristics from a list of predetermined characteristics based on the obtained environmental data.

3. Autonomous navigation method within the meaning of claim 2, where the environmental data one or more include of the following: a camera image, a radar image, a LIDAR image, a ultrasound image, a humidity value, a biofeedback value, a temperature.

4. Autonomous navigation method according to one of the above conclusions, whereby the identifying attributes selecting one or more classes, and optionally subclasses, from a multitude of predefined classes, optionally predefined subclasses, comprises for the classifying characteristics of plants, preferably based on environmental data from one or more sensors.

5. Autonomous navigation method in accordance with one of the above claims, further comprising, after identification of a specific plant, updating the plant data of that plant based of the identified characteristics of that plant.

6. Autonomous navigation method in accordance with one of the above conclusions, furthermore comprising updating the plant data based on one or more models that the development of predict characteristics over time.

7. Autonomous navigation method according to one of the above conclusions, whereby plant characteristics include one or more of the following: - leaf characteristics, including leaf size, leaf patterns, leaf tips, cuts and bruises on the plants, markings, spots, color differences, color contrast; - growth characteristics, including size, number of flowers, fruit and vegetables, growth direction, number leaves, leaf shape, spread, growing point, stem characteristics; - characteristics regarding inanimate objects in the environment, including mounting structures, ground blocks, pipes, wires, beams, floorboards, floor structures, building structures, boards, and markings; - biofeedback characteristics, including sap flow, soil moisture, pest and disease characteristics; - local environmental characteristics, including air temperature, humidity, odor, or sound.

8. Autonomous navigation method in accordance with one of the above conclusions, further comprising: - after retrieving the location of the autonomous object based on one or more identified plants, updating the plant map based on the location of the autonomous object and one or more identified plants.

9. Autonomous navigation method in accordance with claim 8, involving the updating of the plant map furthermore is based on one or more previously retrieved locations of the autonomous object and respectively one or more previously identified plants.

10. Autonomous navigation module for autonomous objects in a controlled agricultural environment, the module comprising: - a memory configured to store a map of the environment, where the map contains locations of predetermined plants, and for storing plant data, whereby the plant data a respective unique combination of plant characteristics with each prior associate a certain plant, - a processing device designed for: or identifying characteristics of a plant in the surrounding environment of the autonomous object, or identifying a specific plant among the predetermined plants based on the identified plant characteristics and the plant data, or retrieving the location of the autonomous object from the map based on the identified plant.

11. Autonomous navigation module pursuant to the previous conclusion, furthermore comprising one or more sensors for obtaining environmental data, where the processing means are further configured to generate identified features from a list of predetermined characteristics based on the obtained environmental data.

12. Autonomous navigation module according to the previous conclusion, where one or more sensors include any of the following: an optical sensor such as a photodetector or an image sensor, a radar, a LIDAR, an ultrasonic sensor, a humidity sensor, a biofeedback sensor, a temperature sensor 13. Autonomous navigation module within the meaning of claim 9 or 10, where the processing means further are configured for generating identified features, furthermore comprising the processing and / or merging acquired environmental data from one or more data sources and / or using a characteristic model based on the obtained environmental data.

14. Autonomous navigation module according to one of the above conclusions, where the memory is configured to store a multitude of predefined classes, optionally predefined certain subclasses, for classifying characteristics of plants, where the processing equipment is further configured for the identification of characteristics by the selecting one or more classes, and optionally subclasses, from the multitude of predefined classes, optional predefined subclasses, preferably based on environmental data from one or more sensors.

15. Autonomous navigation module according to one of the above conclusions, where the processing equipment is further designed for, after identification of a specific plant, updating the plant data of that plant based on the identified characteristics of that plant.

16. Autonomous navigation module according to one of the above conclusions, where the processing equipment is further configured for updating the plant data based on one or more mode models that predict the development of traits over time.

17. Autonomous navigation module according to one of the above conclusions, whereby plant characteristics include one or more of the following: - leaf characteristics, including leaf size, leaf patterns, leaf tips, cuts and bruises on the plants, markings, spots, color differences, color contrast; - growth characteristics, including size, number of flowers, fruit and vegetables, growth direction, number leaves, leaf shape, spread, growing point, stem characteristics; - characteristics related to inanimate objects in the environment, including mounting structures, ground blocks, pipes, wires, beams, floorboards, floor structures, building structures, boards, and markings; - biofeedback characteristics, including sap flow, soil moisture, pest and disease characteristics; - local environmental characteristics, including air temperature, humidity, odor, or sound.

18. Training procedure for training the autonomous navigation module according to one of the the above module conclusions, comprising: - obtaining a training dataset of plant data that contains a respective unique combination associates environmental data with a predetermined characteristic; - training a features model based on the training dataset, to identified to generate characteristics based on acquired environmental data. 94 92 91 200 100 93 100 FIG. 1 200 11 Feature model(s) Plant data Plant map memory 13 12 10 Processor 20 25 Autonomous navigation module 100 Navigation system 35-2 35-1 35-n 40 Sensor 1 Sensor 2 Sensor n Autonomous object 30 30 30 200

Images