Detection and management of target vegetation using machine vision

Patent metadata

Field	Value
Publication number	US-11468670-B2
Application number	US-201816757159-A
Country	US
Kind code	B2
Filing date	Nov 2, 2018
Priority date	Nov 7, 2017
Publication date	Oct 11, 2022
Grant date	Oct 11, 2022

Abstract

Various embodiments detect and manage target vegetation in vegetation areas, including crop beds, between crop beds, and turfgrasses. In one embodiment, a machine learning model is trained to detect target vegetation in captured images. An information processing system is programmed utilizing the machine learning model. One or more images of a particular area are captured, and target vegetation is detected within the one or more images. A position of the detected target vegetation is determined within the one or more images. An applicator disposed on an agrochemical applicator system that is mapped to the position of the detected target vegetation within the one or more images is determined. The applicator is activated based on a current speed of a vehicle coupled to the agrochemical applicator system, and further based on configuration data associated with the applicator. Activating the applicator selectively applies an agrochemical to the detected target vegetation.

First claim

What is claimed is:

1. A detection and management system for detecting and managing target vegetation, the detection and management system comprising:
a moveable entity; and
an agrochemical applicator system mechanically coupled to the moveable entity, the agrochemical applicator system comprising:
a computing device;
an imaging system in data communication with the computing device;
a plurality of applicators for applying an agrochemical to the target vegetation, the plurality of applicators being in data communication with the computing device; and
at least one application executable by the computing device, wherein when executed, the at least one application causes the computing device to at least:
detect the target vegetation within one or more images captured by the imaging system, the target vegetation being detected using a trained deep learning artificial neural network (DLANN) model;
determine a position of the target vegetation within the one or more images; and
comparing the position of the target vegetation with applicator mapping information associated with the plurality of applicators, the applicator mapping information identifying a respective region of the one or more images that is assigned to individual applicators of the plurality of applicators disposed on the agrochemical applicator system;
identifying a particular applicator from the plurality of applicators that is mapped to the position of the target vegetation within the one or more images, the particular applicator being identified based at least in part on the comparison; and
activate the particular applicator according to a current speed of the moveable entity, activating the particular applicator selectively applies the agrochemical to the target vegetation.

2. The detection and management system of claim 1 , further comprising a database, and wherein, when executed, the at least one application further causes the computing device to at least:
store in the database at least one of: a date of an application of the agrochemical to the target vegetation, a location of the application of the agrochemical to the target vegetation, the current speed, the one or more images, an amount of the agrochemical used on the target vegetation, and identified object names within the one or more images.

3. The detection and management system of claim 2 , further comprising a geographic information system (GIS), the GIS being configured to generate one or more maps associated with the target vegetation based at least in part on data stored in the database.

4. The detection and management system of claim 1 , wherein, when executed, the at least one application causes the computing device to at least select the particular applicator from the plurality of applicators based at least in part on a location of the target vegetation and a position of the particular applicator, activating the particular applicator.

5. The detection and management system of claim 1 , wherein determining the position of the target vegetation is based at least in part on a location determining system data and pixel coordinates of the target vegetation.

6. The detection and management system of claim 1 , wherein the agrochemical applicator system further comprises one or more tanks holding the agrochemical.

7. The detection and management system of claim 1 , wherein the one or more images are of a vegetation area comprising at least one of: a crop bed, an area between two crop beds, or turfgrass.

8. A method for detecting and managing target vegetation within a vegetation area, the method comprising:
capturing one or more images of the vegetation area via an imaging device;
detecting the target vegetation within the one or more images using a neural network model;
determining a position of the target vegetation within the one or more images;
identifying a particular applicator of a plurality of applicators disposed on an agrochemical applicator system, the particular applicator being identified based at least in part on a comparison of the position of the target vegetation with applicator mapping information associated with the plurality of applicators, the applicator mapping information identifying a respective region of the one or more images that is assigned to individual applicators of the plurality of applicators, and the particular applicator being mapped to the position of the target vegetation; and
activating the particular applicator based on a current speed of a vehicle coupled to the agrochemical applicator system, wherein activating the particular applicator selectively applies an agrochemical to the target vegetation.

9. The method of claim 8 , wherein the vegetation area comprises at least one of turfgrass, a crop bed, or an area between crop beds.

10. The method of claim 8 , further comprising training the neural network model to detect the target vegetation within the one or more images.

11. The method of claim 10 , further comprising:
programming an information processing system utilizing the the neural network model; and
wherein the information processing system comprises one or more object detectors trained to identify the target vegetation within the one or more images and detect the target vegetation within the one or more images.

12. The method of claim 8 , wherein activating the particular applicator is further based at least in part on configuration data associated with the particular applicator.

13. The method of claim 12 , wherein the configuration data comprises at least one of: a position of the particular applicator relative to a position of the imaging device, a distance between the particular applicator and the imaging device, a dispensing angle of the particular applicator, a coverage area of the particular applicator, an actuation speed of the particular applicator, a field-of-view of the imaging device, a distance of the particular applicator from a ground, a location of the particular applicator on the agrochemical applicator system, or a mapping of the particular applicator to a portion of the field-of-view of the imaging device.

14. The method of claim 13 , wherein the agrochemical applicator system comprises a motorized track, the particular applicator is coupled to the motorized track, and the configuration data further comprises at least one of: a speed of travel of the motorized track coupled to the particular applicator or a current position of the particular applicator on the motorized track.

15. The method of claim 8 , wherein the agrochemical applicator system comprises a global positioning system (GPS) module, and further comprising determining the current speed of the vehicle via data obtained from the GPS module.

16. An agrochemical applicator system for detecting target vegetation, the agrochemical applicator system comprises:
one or more applicators configured to dispense an agrochemical;
an imaging system comprising at least one imaging device; and
at least one application executable in a computing device, wherein, when executed, the at least one application causes the computing device to at least:
detect the target vegetation within one or more images utilizing a trained neural network model, the one or more images being captured by the imaging system;
determine a position of the target vegetation within the one or more images;
determine that a particular applicator from the one or more applicators is mapped to the position of the target vegetation within the one or more images based at least in part on a comparison of the position of the target vegetation with applicator mapping information a

Assignees

• Univ Florida

Inventors

• Schumann Arnold W
• Boyd Nathan S
• Yu Jialin

Classifications

• G06N20/00

  Machine learning · CPC title

• G06V20/188 Primary

  Vegetation · CPC title

• A01M7/0089 Primary

  Regulating or controlling systems (the delivery being related to the movement of a vehicle B05B9/06) · CPC title

• G06T7/70

  Determining position or orientation of objects or cameras (camera calibration G06T7/80) · CPC title

• G06T2207/30188

  Vegetation; Agriculture · CPC title