Robotic Toxicodendron weeding

Patent metadata

Field	Value
Publication number	US-11783207-B2
Application number	US-202016793792-A
Country	US
Kind code	B2
Filing date	Feb 18, 2020
Priority date	Feb 18, 2020
Publication date	Oct 10, 2023
Grant date	Oct 10, 2023

Abstract

A system includes a memory having instructions therein and at least one processor in communication with the memory. The at least one processor is configured to execute the instructions to acquire phytomorphological field data via a sensor component of a mobile robot, generate, based on the phytomorphological field data and via a machine learning agent, a predicted likelihood of whether a hypothetical action by the mobile robot against a found plant would be directed against a true Toxicodendron plant, conduct a non-phytomorphological assessment of the found plant via the mobile robot and based on the predicted likelihood being below a first threshold and above a second threshold, and, via the mobile robot and based on the non-phytomorphological assessment, attack the found plant, mark a site of the found plant, and/or document a context of the site.

First claim

What is claimed is:

1. A method, comprising:
acquiring phytomorphological field data (PFD) using a sensor component of a mobile robot;
receiving a likelihood assumption threshold (LAT) related to conducting a non-phytomorphological assessment;
receiving a likelihood confirmation threshold (LCT) related to conducting a non-phytomorphological assessment,
receiving a confirmation assessment threshold (CAT) related to a non-phytomorphological assessment;
generating, based on the PFD and using a machine learning agent, a PFD predicted likelihood of whether a hypothetical action by the mobile robot against a found plant would be directed against a true Toxicodendron plant;
when the PFD predicted likelihood is less than or equal to the LAT threshold and the PFD predicted likelihood is greater than the LCT threshold;
conducting a non-phytomorphological assessment of the found plant using the mobile robot,
generating a non-phytomorphological confidence score (NCS) based on the non-phytomorphological assessment, and
when the NCS is greater than the CAT, taking at least one actual action; and

when the PFD predicted likelihood is greater than the LAT threshold, taking the at least one actual action,
wherein the at least one actual action is selected from the group consisting of attacking the found plant, marking a site of the found plant, and documenting a context of the site, and
wherein the action is taken using the mobile robot.

2. The method of claim 1 , further comprising:
training the machine learning agent using phytomorphological training data; and
automatically initiating retraining of the machine learning agent based on the non-phytomorphological assessment,
wherein the phytomorphological training data excludes the phytomorphological field data and the retraining of the machine learning agent comprises using the phytomorphological field data.

3. The method of claim 1 , wherein the sensor component comprises a camera and the phytomorphological field data comprises image data.

4. The method of claim 3 , wherein conducting the non-phytomorphological assessment of the found plant comprises detecting a plant exudate.

5. The method of claim 4 , wherein detecting the plant exudate comprises applying a reagent to the plant exudate.

6. The method of claim 5 , wherein detecting the plant exudate comprises detecting urushiol.

7. The method of claim 6 , wherein detecting the plant exudate comprises:
applying, using an emitter component of the mobile robot, ultraviolet radiation to a product of a reaction of the reagent with the plant exudate.

8. A system, comprising:
a memory having instructions therein; and
at least one processor in communication with the memory, wherein the at least one processor is configured to execute the instructions to:
acquire phytomorphological field data (PFD) via a sensor component of a mobile robot;
receive a likelihood assumption threshold (LAT) related to conducting a non-phytomorphological assessment;
receive a likelihood confirmation threshold (LCT) related to conducting a non-phytomorphological assessment:
receive a confirmation assessment threshold (CAT) related to a non-phytomorphological assessment;
generate, based on the PFD and via a machine learning agent, a PFD predicted likelihood of whether a hypothetical action by the mobile robot against a found plant would be directed against a true Toxicodendron plant;
when the PFD predicted likelihood is less than or equal to the LAT threshold and the PFD predicted likelihood is greater than the LCT threshold:
conduct a non-phytomorphological assessment of the found plant via the mobile robot,
generate a non-phytomorphological confidence score (NCS) based on the non-phytomorphological assessment, and
when the NCS is greater than the CAT, take at least one actual action; and

when the PFD predicted likelihood is greater than the LAT threshold, take the at least one actual action,

wherein the at least one actual action is selected from the group consisting of an attack on the found plant, a marking of a site of the found plant, and a documenting of a context of the site, and
wherein the action is taken via the mobile robot.

9. The system of claim 8 , wherein the at least one processor is further configured to execute the instructions to:
train the machine learning agent with phytomorphological training data; and
automatically initiate a retraining of the machine learning agent based on the non-phytomorphological assessment,
wherein the phytomorphological training data excludes the phytomorphological field data and the retraining of the machine learning agent comprises use of the phytomorphological field data.

10. The system of claim 8 , wherein the sensor component comprises a camera and the phytomorphological field data comprises image data.

11. The system of claim 10 , wherein the at least one processor is further configured to execute the instructions to use the mobile robot to detect a plant exudate.

12. The system of claim 11 , wherein the at least one processor is further configured to execute the instructions to use the mobile robot to apply a reagent to the plant exudate.

13. The system of claim 12 , wherein the at least one processor is further configured to execute the instructions to use the mobile robot to detect urushiol.

14. The system of claim 13 , wherein the at least one processor is further configured to execute the instructions to use the mobile robot to apply ultraviolet radiation to a product of a reaction of the reagent with the plant exudate.

15. A computer program product, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by at least one processor to cause the at least one processor to:
acquire phytomorphological field data (PFD) via a sensor component of a mobile robot;
receive a likelihood assumption threshold (LAT) related to conducting a non-phytomorphological assessment;
receive a likelihood confirmation threshold (LCT) related to conducting a non-phytomorphological assessment;
receive confirmation assessment threshold (CAT) related to a non-phytomorphological assessment;
generate, via a machine learning agent and based on the PFD, a PFD predicted likelihood of whether a hypothetical action by the mobile robot against a found plant would be directed against a true Toxicodendron plant;

when the PFD predicted likelihood is less than or equal to the LAT threshold and the PFD predicted likelihood is greater than the LCT threshold;
use the mobile robot to conduct a non-phytomorphological assessment of the found plan, and
generate a non-phytomorphological confidence score (NCS) based on the non-phytomorphological assessment, and
when the NCS is greater than the CAT, take at least one actual action; and

when the PFD predicted likelihood is greater than the LAT threshold, take the at least one actual action,
wherein the at least one actual action is selected from the group consisting of an attack on the found plant, a marking of a site of the found plant, and a documenting of a context of the site, and
wherein the at least one actual action is taken via the mobile robot.

16. The computer program product of claim 15 , wherein the program instructions are further executable by the at least one processor to cause the at least one processor to:
train the machine learning agent with phytomorphological training data; and
automatically initiate a retraining of the machine learning agent based on the non-phytomorphological assessment,

Assignees

• IBM

Inventors

• Emanuel Barton Wayne
• Kochura Nadiya
• Suto Tiberiu
• Valecha Vinod A

Classifications

• G06N5/04 Primary

  Inference or reasoning models · CPC title

• A01M21/02

  Apparatus for mechanical destruction {(hand-tools for uprooting weeds A01B1/16; machines for working soil and for weeding A01B39/18; harvesting machines A01D; cutting implements A01G3/00)} · CPC title

• B25J9/163 Primary

  learning, adaptive, model based, rule based expert control · CPC title

• G05B13/0265

  the criterion being a learning criterion · CPC title

• G06N20/00

  Machine learning · CPC title