Robotic cultivator

• US 5,442,552 A
• Filed: 03/16/1993
• Issued: 08/15/1995
• Est. Priority Date: 03/16/1993
• Status: Expired due to Fees

• Current Assignee: Regents of the University of California (University of California)
• Original Assignee: Regents of the University of California (University of California)
• Inventors: Chen, Pictiaw, Curley, Robert G., Slaughter, David C., Giles, Durham K.
• Primary Examiner(s): Powell, Mark R.
• Assistant Examiner(s): SAX, STEVEN PAUL
• Application Number:
  US08/032,075
• Time in Patent Office: 882 Days
• Field of Search: 364/424.07, 364/413.13, 382/10, 382/11, 382/12, 382/13, 382/14, 382/15, 358/455
• US Class Current: 701/50
• CPC Class Codes: G05D 1/0246 using a video camera in combination with image processing means
• CPC Class Codes: G06V 10/25 Determination of region of interest [ROI] or a volume of interest [VOI]

Abstract

A robotic cultivator is provided for automatically positioning cultivation tools relative to a centerline of a plant row. The cultivator comprises a video camera for gathering visual information about plants in a plant row. A frame grabber is then utilized to generate a digitized image of the visual information. A computer then manipulates the visual information to determine a centerline of the plant row. Finally, a positioning device moves the cultivation tools based upon the current position of the tools with respect to the centerline.

Claims

1. A robotic cultivator for automatically positioning cultivation tools relative to a centerline of a plant row, said cultivator comprising:

visual input means for gathering visual information about plants and weeds in said plant row;

digitizing means for generating a digitized image of said visual information;

distinguishing means for distinguishing between said plants and said weeds in said digitized image;

means for determining a centerline of only said plants based upon said digitized image of said visual information; and

means responsive to said means for determining a centerline for locating said tools relative to said centerline.

2. The cultivator recited in claim 1, wherein said visual input means comprises a video camera.

3. The cultivator recited in claim 1, wherein said digitizing means comprises a frame grabber.

4. The cultivator recited in claim 1, wherein said means responsive to said means for determining a centerline comprises:

an asynchronous receiver and transmitter;

an actuator for actuating a displacement device attached to said tool;

a locating means for determining the position of said tool; and

a central processing unit for receiving position information from said locating means and for actuating said actuator to move said tool to said centerline.

5. The cultivator recited in claim 4, wherein said locating means comprises a linear potentiometer.

6. The cultivator recited in claim 4, further comprising a manual adjusting means.

7. The cultivator recited in claim 6, wherein said manual adjusting means comprises at least one joystick.

8. The cultivator recited in claim 2, wherein said camera is mounted so as to have a centerline of said camera directly above a centerline of a pair of said tools.

9. The cultivator recited in claim 2, wherein said camera is orientated so that a long aspect ratio of said camera is parallel to said plant row.

10. The cultivator recited in claim 1, further comprising at least one band spray nozzle, said nozzle being rotatable about a vertical axis thus allowing a spray from said nozzle to be adjusted with respect to the direction of travel of said cultivator.

11. A method of controlling a robotic cultivator, having tools thereon, based upon visual information of plants in a plant row, said method comprising the steps of:

generating a visual image of plants and weeds in said plant row, said visual image being represented by pixels;

digitizing said pixels based upon visual characteristics of said plants and weeds;

generating mean and covariance values based upon said visual characteristics;

generating a probability table from said mean and covariance values;

comparing each of said pixels with said probability table to determine whether each of said pixels represents a plant;

generating a representative image from said digitized image where a logic state of one is assigned to each pixel representing a plant; and

determining a centerline of only said plants based upon said representative image.

12. The method recited in claim 11, wherein said step of generating a probability table comprises the steps of:

using Bayes Law to compute the probability that a particular color will be in a plant; and

assigning the value of a logical one to all probabilities where a particular color will be in said plant.

13. The method recited in claim 11, wherein said step of determining a centerline comprises the steps of:

scanning each pixel in said representative image, one row at a time, to determine if the value of that particular pixel is non zero;

incrementing one of a plurality of counters, said particular incremented counter associated with said pixel, if said pixel has a non zero value;

calculating a histogram from said plurality of counters;

calculating a mean of said histogram; and

assigning said centerline to the location corresponding to a mean of said histogram.

14. The method recited in claim 13, further comprising the steps of:

dividing said plurality of counters into subsets;

calculating a histogram of each subset;

calculating a median for each histogram;

disregarding any median which significantly differs from an average of said medians of each histogram; and

assigning said centerline location based upon the remaining medians and their corresponding physical locations.

15. The method recited in claim 11, wherein said step of determining a centerline comprises the steps of:

dividing said representative image into a plurality of regions with associated counters;

scanning each pixel in said representative image to determine if the value of that particular pixel is non zero;

determining which region each pixel having a non zero value is located and incrementing said associated counter;

repeating said scanning and determining steps until each pixel has been scanned;

comparing each counter to determine which counter has the maximum value; and

assigning the location corresponding to the counter with the maximum value to said centerline.

16. The method recited in claim 11, further comprising the steps of:

determining a position of said tools;

comparing said tool position to a position relative to said centerline to generate an position error signal; and

moving said tools to reduce said error signal;

repeating said determining and comparing steps until said error signal is within a predetermined tolerance.