Boundary detecting method and apparatus for automatic working vehicle

• US 4,868,752 A
• Filed: 09/10/1987
• Issued: 09/19/1989
• Est. Priority Date: 07/30/1987
• Status: Expired due to Fees

• Current Assignee: Kubota Corporation
• Original Assignee: Kubota Corporation
• Inventors: Hayashi, Masahiko, Fujii, Yasuo
• Primary Examiner(s): James, Andrew J.
• Assistant Examiner(s): Soltz, David
• Application Number:
  US07/094,996
• Time in Patent Office: 740 Days
• Field of Search: 364/424, 364/424.02, 364/424.07, 382/41
• US Class Current: 701/28
• CPC Class Codes: G05D 1/0246 using a video camera in combination with image processing means

Abstract

A method and apparatus for use with an automatic working vehicle for detecting a boundary between a treated area and an untreated area. A predetermined range of working area including the treated area and the untreated area is photographed by a camera. Image data output from the camera is differentiated with respect to pixel to pixel brightness variations, which is then compared with a threshold value to produce edge image data on an X-Y plane showing, as binarized, a plurality of pixels regarded as representing a point of transition between the untreated area and the treated area, and other pixels. The boundary is calculated from the edge image data by means of HOUGH-transformation.

Claims

1. A boundary detecting method for an automatic working vehicle, comprising the steps of:

picking up an image (SG) of a working area in a predetermined range forwardly of the vehicle and converting said image into digital data (F1);

converting data (F1) of the image picked up into data (F2) differentiated with respect to pixel to pixel brightness variations;

comparing the differentiated data (F2) with a threshold value to produce binarized edge image data (F3) on an X-Y plane showing, as binarized, a plurality of pixels regarded as representing a point of transition between an untreated area (B) and a treated area (C) and other pixels;

providing a lookup table for precalculating and storing a group of HOUGH-transformation coordinate points [ρ

, θ

] corresponding to a group of straight lines, said straight lines further corresponding to the number of dispersed points defined on the ρ

-θ

plane passing through each coordinate point [x, y] on the X-Y plane;

calculating a boundary between the untreated area (B) and the treated area (C) from the binarized edge image data (F3) by means of HOUGH-transformation, a ρ

-θ

plane in HOUGH-transformation being dispersed to define points on a grid; and

the step of calculating a boundary including the further steps of reading the group of HOUGH-transformation corresponding coordinate points in response to respective transition point coordinates in the edge image data from the lookup table, counting frequencies with which the points on the grid on the ρ

-θ

plane in the HOUGH-transformation are read, and scanning said frequencies and determining a maximum value of the frequencies so as to detect a straight line corresponding to the maximum read point, thereby determining a boundary.

2. A boundary detecting apparatus for an automatic working vehicle, comprising:

image pickup means for detecting an image (SG) of a working area in a predetermined range forwardly of the vehicle;

A/D converter means for converting said image (SG) into digital image data (F1);

differentiating means for converting image data (F1) received from the image pickup means into image data (F2) differentiated with respect to pixel to pixel brightness variations;

binarizing means for comparing the differentiated data (F2) with a threshold value and producing binarized edge image data (F3) on an X-Y plane showing, as binarized, a plurality of pixels regarded as representing a point of transition between an untreated area (B) and a treated area (C) and other pixels; and

HOUGH-transformation means for receiving the edge image data (F3) and calculating a boundary between the untreated area (B) and the treated area (C) from the edge image data (F3) by means of HOUGH-transformation, the HOUGH-transformation means including;

a lookup table for dispersing a ρ

-θ

plane in HOUGH-transformation, defining points on a grid and precalculating and storing a group of HOUGH-transformation coordinate points [ρ

, θ

] corresponding to a group of straight lines which further correspond to the number of the dispersed θ

on the ρ

-θ

plane passing through each coordinate [x, y] on the X-Y plane;

means for reading the group of HOUGH-transformation corresponding coordinate points in response to the respective transition point coordinates in the edge image data from the lookup table;

means for counting frequencies with which the points on the grid on the ρ

-θ

plane in the HOUGH-transformation are read; and

means for scanning and determining a maximum value of the frequencies so as to detect a straight line corresponding to the maximum read point and determine a boundary.

3. A boundary detecting apparatus as claimed in claim 2 wherein the lookup table constitutes a ROM including addresses consisting of the coordinates of points [x, y] on the X-Y plane and one of the mapped coordinate values on the HOUGH-transformation plane associated with the respective coordinates [x, y] and data consisting of the other mapped coordinate value.

4. A boundary detecting apparatus as claimed in claim 2, further comprising:

coordinate turning means disposed at an input end of the A/D converter means for rotating by 90 degrees the coordinates of the image data (SG) received from the image pickup means and for selectively outputting results to the differentiating means; and

image switching means for selectively receiving the image data (SG) transmitted from the image pickup means bypassing the memory means and for directly transmitting the digital image data (F1) to the differentiating means.

5. A boundary detecting apparatus as claimed in claim 2, wherein the means for reading the group of the HOUGH-transformation corresponding coordinate points [ρ

, η

] from the lookup table and counting frequencies with which the points in the grid on the ρ

-θ

plane in the HOUGH-transformation are read is adapted to select from points within a narrow scope of zero to 180 degrees which does not substantially vary from a forward direction of the vehicle (V).

6. An automatic working vehicle having a boundary detecting apparatus, comprising:

steerable wheels;

steering control means for controlling the steerable wheels to cause the working vehicle to automatically run along a boundary between an untreated area (B) and a treated area (C); and

boundary detecting means for detecting said boundary, said boundary detecting means including;

image pickup means for detecting an image of a working area in a predetermined range forwardly of the vehicle;

A/D converter means for converting said detected image to digital image data (F1);

differentiating means for converting image data (F1) received from the image pickup means into data (F2) differentiated with respect to pixel to pixel brightness variations;

binarizing means for comparing the differentiated data (F2) with a threshold value and producing binarized edge image data (F3) on an X-Y plane showing, as binarized, a plurality of pixels regarded as representing a point of transition between the untreated area (B) and the treated area (C) and other pixels;

HOUGH-transformation means for receiving the edge image data (F3) and calculating the boundary between the untreated area (B) and the treated area (C) from the edge image data (F3) by means of HOUGH-transformation; and

control means for outputting to the steering control means a control signal corresponding to the boundary calculated by the HOUGH-transformation means, wherein the HOUGH-transformation means includes;

a lookup table for dispersing a ρ

-θ

plane in HOUGH-transformation, defining points on a grid, and precalculating and storing a group of HOUGH-transformation coordinate points [ρ

, θ

] corresponding to a group of straight lines which correspond to the number of the dispersed θ

on the ρ

-θ

plane passing through each coordinate [x, y] on the X-Y plane;

means for counting frequencies with which the points on the grid on the ρ

-θ

plane in the HOUGH-transformation are read;

means for scanning and determining a maximum value of the frequencies so as to detect a straight line corresponding to the maximum read point and determine a boundary; and

control means for sending a control signal corresponding to the boundary calculated by the HOUGH-transformation means to the steering control means.

7. An automatic working vehicle as claimed in claim 6, wherein the steering control means is operable to control the steerable wheels to follow the boundary determined by the HOUGH-transformation means to be represented by at least one position in a central region transversely of a photographic field of vision of the image pickup means.

8. An automatic working vehicle as claimed in claim 7, wherein the image pickup means is attached to the vehicle with a photographing direction thereof extending obliquely downward.

9. An automatic working vehicle as claimed in claim 8, wherein the image pickup means includes an optical filter for shading the photographic field of vision to a greater degree progressively toward a portion thereof corresponding to a rear end longitudinally of the vehicle.