CN100349689C - Automatic weld seam position based on circular laser vision sense - Google Patents

Abstract

The present invention relates to an automatic positioning method for a welding seam based on annular laser visual sensing, which belongs to the technical field of welding. The present invention comprises the following steps: (1), the representation equations of Y-coordinates and X-coordinates of a characteristic point of the welding seam in a laser conical coordinate system are acquired in a coordinate system of an image sensing surface according to an image of an area array CCD detection annular laser detection track on the image sensing surface of a video camera; (2), the algorithm of the relation of the radius and the depth of a laser conical coordinate system is used for solving the depth value of the characteristic point of the welding seam, and an equation of a Z-coordinate of a detected point based on the characteristic point of the welding seam in the laser conical coordinate system is built and solved. (3), based on the matrix conversion of the laser conical coordinate system and a robot body, solved three-dimensional coordinates in the laser conical coordinate system are converted to a coordinate system of a robot body, and therefore, the positioning to the welding seams is realized. The present invention solves the problem of low efficiency due to the fact that linear array CCD visual sensors based on point-shaped laser, linear laser and annular laser use a triangulation principle to position the welding seams.

Description

Automatic weld seam position based on circular laser vision sense

Technical field

What the present invention relates to is a kind of method of welding technology field, specifically, is a kind of automatic weld seam position based on circular laser vision sense.

Background technology

Robot and computer technology rapid development improve the Automation of Welding degree day by day.The factor that influences automation welding postwelding quality not only comprises the internal factor of the welding such as chemical composition of the thickness of welded piece, surface treatment, welding material, also comprises important informations such as welding method, speed of welding, welding torch path planning, location of workpiece detection.Particularly the three-dimensional position of welded piece detects, and is the major issue that Automation of Welding faced always.Welding process rises to the corresponding development that automation and artificial intelligence level need the identification of weld seam information, location and tracking technique.And not accurately, efficient and react sensor fast, the target of Automation of Welding just can't realize.In order to realize Automation of Welding, automatically obtain the locating information of weld seam in real time, developed many sensing technologies, wherein, vision sensor is owing to its noncontact, quick, the strong extensive concern that is subjected to of antijamming capability of reaction.Initiatively the light vision technology utilizes the method for structured light method or optical scanner to produce certain laser trace and with its detection means as weld seam.Detection laser track commonly used has: point-like laser, single line bar laser, two-wire bar laser, five lines laser and loop lasers, but point-like or linear laser vision sensor information content is few, detection efficiency is low; Exist explanation fuzzy; The tracking accuracy difference reaches the location and the following function that only can realize weld seam on one dimension direction or the two-dimensional directional.Depth computing method is a principle of triangulation.Utilizing line array CCD also is to utilize triangulation method, claimed structure to meet the Scheimpflug condition as the circular laser vision sense device of recipient, and it is low to measure efficient.

Find through literature search prior art, Wu Minsheng rolls up the P333-339 page or leaf in " Welding in the World " (" the welding world ") September the 34th in 1994 and has published " New scanning method of opticalsensor system for seam tracking in three dimensions is used for the three-dimensional novel scan method of following the tracks of optical sensor system of a weld seam " literary composition, proposes a kind of continuous circular shape optical mechaical scanning method that is used for weld joint tracking in the literary composition.Its implementation optical texture complexity, and need to obtain analog signal through the secondary opto-electronic conversion, the stepper motor driven gear is secondary to be realized by driving in the rotation of semiconductor light sources, precision is low, shuttle-scanning speed descends, mechanical oscillation are big, three-dimensional information is tried to achieve for approximate, its depth extraction method has utilized principle of triangulation, and detection efficiency is low.

Summary of the invention

The object of the invention is to overcome deficiency of the prior art, a kind of automatic weld seam position based on circular laser vision sense is provided, the problem that makes its depth information of not only avoiding traditional passive light vision sensor weld seam be difficult to determine has also solved based on point-like laser, linear laser and line array CCD vision sensor and has utilized principle of triangulation to carry out the inefficient problem that the weld seam location is caused.

The present invention is achieved by the following technical solutions, the present invention includes following steps:

(1) laser diode emission point-like laser produces loop laser through relay lens, and is imaged on the video camera image sensing surface, obtains loop laser by area array CCD and detects track.In the image sensing surface coordinate system, by the imaging acquisition Y coordinate of weld bead feature points in laser cone coordinate system and the expression equation of X coordinate of video camera;

Described by area array CCD acquisition loop laser detection track, its method is to be stored by area array CCD by the signal charge that loop laser produces, and converts the 2-D data information of reflection loop laser to.

The imaging of described video camera, its method is that detected loop laser is mapped on the light-sensitive array of Array CCD sensor through optical lens, by CCD optical signal is converted into the signal of telecommunication, when the sequential generator applies the pulse of specific time sequence to it, the charge signal of each pixel will shift out CCD and successively through zooming into the analog signal that voltage amplitude does not wait, these signals are sent to independently analog front circuit and convert data signal to of other cities, and these data signals are exactly the original image information that sensor captures.

(2) utilize radius of circle and depth relationship algorithm depth value with weld bead feature points, it is the Z coordinate in the laser cone coordinate system, radius of circle with the standard of demarcating is represented, structure is based on the equation of the X coordinate of weld bead feature points under laser cone coordinate system and find the solution, obtain the X coordinate, and then try to achieve the three-dimensional coordinate of weld bead feature points;

Described radius of circle and depth relationship algorithm are meant the depth z of tested measuring point _LRepresent with radius r, promptly $z_L=\frac{175(r_0-r)}{r_0},$ Wherein, r ₀-standard laser track radius of a circle is a constant; 175-relay lens focal length.

(3) according to the matrix conversion of laser cone coordinate system and robot body coordinate system, the three-dimensional coordinate in the laser cone coordinate system of trying to achieve is transformed into the robot body coordinate system, thereby realizes the location of butt welded seam.

Described three-dimensional coordinate is transformed into the robot body coordinate system, its method is to demarcate by the arm end of robot, determine the matrix relationship between the terminal coordinate system of robot body coordinate system and welding torch, by the matrix relationship between terminal coordinate system of the welding torch of demarcating and the laser cone coordinate system, determine to be transformed into the matrix relationship of robot body coordinate system again from laser cone coordinate system.

The present invention utilizes loop laser as weld seam information sensing means.The point-like laser of laser diode emission projects on the welded piece to be detected ringwise through after the rotation sweep effect of vision sensor, and utilizes area array CCD to receive.After the characteristic point process image filtering of position while welding, the image dividing processing, the pixel of circular laser trace in area array CCD carried out the characteristic point that difference algorithm identifies weld seam.Utilize " radius of circle and the depth relationship " algorithm that proposes to obtain the three-dimensional information of weld bead feature points.Obtain to represent the coordinate of characteristic point in the robot body coordinate system of position while welding.

The present invention utilizes the visual sensing of loop laser as the weld seam localization method, because vision sensor of the present invention has adopted the optical scanner method, avoided the significantly shortcoming of beam split of structured light method lasing light emitter energy, thereby strengthened the operability of weld seam recognition and location, complex structure, inefficient problem when also having avoided simultaneously loop laser line array CCD visual sensing method to be used for weld seam recognition with the location have guaranteed the real-time detection of weld seam location.The more important thing is, the visual sensing method of present existing point-like laser, linear laser and loop laser is to utilize principle of triangulation to realize that the weld seam three-dimensional coordinate calculates, the inventive method utilizes " radius of circle and depth relationship " algorithm to calculate the three-dimensional coordinate of weld bead feature points, has higher efficient and precision.Be applicable to the real-time location of the weld seam in the automatic welding process of robot, also can measure welding point parameter commonly used simultaneously, as the gap length of banjo fixing butt jointing, the width of weld seam and the degree of depth etc.

Description of drawings

Fig. 1 is the three-dimensional coordinate schematic diagram calculation.

Fig. 2 is weld bead feature points recognition principle figure.

The specific embodiment

As shown in Figure 1, the point-like laser of laser diode emission projects detected welded piece surface by the scan action of the relay lens of laser vision sensor.Array CCD Camera detects weld seam and circular laser trace by image pick-up card, realize the A/D conversion and collect computer, computer utilizes difference algorithm to identify the characteristic point that the laser scanning track detection arrives after image such as cutting apart and handle by image filtering, image again.These characteristic points are the position at weld seam place.At last, obtain the three-dimensional coordinate of weld bead feature points in laser cone coordinate system according to " radius of circle and the depth relationship " algorithm that proposes.And the transformational relation by laser cone coordinate system and robot world's coordinate system, the three-dimensional coordinate of weld bead feature points in laser cone coordinate system is converted to the three-dimensional coordinate of robot body coordinate system, detected weld seam physical location passed to robot controller by the communication between computer and the robot.Thereby make robot obtain the three-dimensional location coordinates of weld seam in real time.

As shown in Figure 2.The recognizer of weld bead feature points is based on X, the difference algorithm of Y coordinate.Computation model is as follows: establish x in the camera pixel coordinate, y is respectively the function of off-axis angle, then gradient function G _x' (γ), G _y' (γ) be respectively:

$G_x^{\′}\left(\mathrm{\γ}\right)=\frac{\mathrm{dx}}{\mathrm{d\γ}}$

$G_y^{\′}\left(\mathrm{\γ}\right)=\frac{\mathrm{dy}}{\mathrm{d\γ}}$

(1)

But utilize the difference algorithm evaluation at γ=a point:

$x^{\′}\left(a\right)\≈\frac{x(a+h)-x\left(a\right)}{h}$

$y^{\′}\left(a\right)\≈\frac{y(a+h)-y\left(a\right)}{h}$

(2)

Wherein, h is a step-length.

If γ _k=γ ₀+ kh (k=0,1,2 ..., n), then backward difference operator  can describe with following formula:

x _k＝x _k-x _k-1

y _k＝y _k-y _k-1 (3)

Characteristics according to the lap joint of Fig. 2 exist breakpoint in the weld bead feature points position, provide recognition strategy.

The weld seam recognition strategy:

If  is y _k＞=2,  x _k＞5 these points belong to weld bead feature points.

As shown in Figure 2, for the arbitrfary point P (x on the weld seam _L, y _L, z _L) available this two-dimensional coordinate P ' (x in camera coordinate system _c, y _c) represent:

$\left\{\begin{array}{c}{x}_L-z_L\tan \mathrm{\β}=k_x\·x_C\\ y_L=k_y\·y_C\end{array}\right.---\left(4\right)$

In laser cone coordinate system, the relation of the radius of a circle and the degree of depth on the laser scanning conical surface (Z coordinate figure), x _L, y _LAnd there is following relation between the radius r:

(x _L-r) ²+y _L ²＝r ² (5)

Order

a＝2r ₀+175tanβ

b＝2·r ₀(x _c-175tanβ)

c＝175y _c ²tanβ (6)

So have,

ax _L ²+bx _L+c＝0 (7)

Solve P (x _L, y _L, z _L) coordinate is

$x_L=\frac{-b\±\sqrt{b^2-4\·a\·c}}{2\·a}$

y _L＝k _y·y _C

$z_L=\frac{175(r_0-r)}{r_0}$

(8)

Utilize " radius of circle and depth relationship " algorithm of formula (4), (5) and (8), can calculate the weld bead feature points three-dimensional value.

As shown in Figure 2, obtain to describe the characteristic point of weld seam according to difference algorithm: A, B, C and D.O and O ' is the original point of weld seam.Connect the power supply of video camera, motor and laser diode, the circular laser vision sense device projects laser on the weld seam of lap joint, as shown in Figure 2, carries out the A/D conversion and computer is arrived in the information gathering of weld seam position by image pick-up card.To image carry out filtering, cut apart with thinning processing after, utilize difference algorithm identification to describe the characteristic point of weld seam: A, B, C and D.According to the pixel coordinate of these points and " radius of circle and depth relationship " algorithm of proposition, obtain its coordinate in laser cone coordinate system and be: A (37.2781-2.319 12.4373); B (0.2072-2.7828 11.6472); C (38.0022-3.0147 8.4853); D (0.4076-3.9422 7.8541).So the position while welding that identifies is: O (37.6402-2.6668 10.4613); O ' (0.3074-3.3625 9.3507), the position while welding that O and O ' expression vision sensor identify, the direction to be welded of vector OO ' expression weld seam.

Claims (5)

1. the automatic weld seam position based on circular laser vision sense is characterized in that, may further comprise the steps:

(1) laser diode emission point-like laser, produce loop laser through relay lens, project imaging on the weld seam, detect loop laser inspection mark by area array CCD, in the image sensing surface coordinate system, by the imaging acquisition Y coordinate of weld bead feature points in laser cone coordinate system and the expression equation of X coordinate of video camera;

(2) utilize in the laser cone coordinate system radius of circle and depth relationship algorithm with the depth value of weld bead feature points, it is the Z coordinate in the laser cone coordinate system, radius of circle with the standard of demarcating is represented, structure is based on the equation of the Z coordinate of weld bead feature points under laser cone coordinate system and find the solution, obtain the Z coordinate, and then try to achieve the three-dimensional coordinate of weld bead feature points;

(3) according to the matrix conversion of laser cone coordinate system and robot body coordinate system, the three-dimensional coordinate in the laser cone coordinate system of trying to achieve is transformed into the robot body coordinate system, thereby realizes the location of butt welded seam.

2. the automatic weld seam position based on circular laser vision sense according to claim 1, it is characterized in that, in the step (1), described method by area array CCD acquisition loop laser detection track is to be stored by area array CCD by the signal charge that loop laser produces, and converts the 2-D data information of reflection loop laser to.

3. the automatic weld seam position based on circular laser vision sense according to claim 2, it is characterized in that, in the step (1), the method of described video camera imaging is that detected loop laser is mapped on the light-sensitive array of Array CCD sensor through optical lens, by CCD optical signal is converted into the signal of telecommunication, when the sequential generator applies the pulse of specific time sequence to it, the charge signal of each pixel will shift out CCD and successively through zooming into the analog signal that voltage amplitude does not wait, these signals are sent to independently analog front circuit and convert data signal to of other cities, and these data signals are exactly the original image information that sensor captures.

4. the automatic weld seam position based on circular laser vision sense according to claim 1 is characterized in that, described radius of circle and depth relationship algorithm are meant the depth z of tested measuring point _LRepresent with radius r, promptly $z_L=\frac{175(r_0-r)}{r_0},$ R wherein ₀Being the standard radius of a circle, is a constant; 175-focal length of lens f.

5. the automatic weld seam position based on circular laser vision sense according to claim 1, it is characterized in that, described three-dimensional coordinate is transformed into the robot body coordinate system, its method is the demarcation by robot, determine the matrix relationship between the terminal coordinate system of robot body coordinate system and welding torch, by the matrix relationship between terminal coordinate system of the welding torch of demarcating and the laser cone coordinate system, determine to be transformed into the matrix relationship of robot body coordinate system again from laser cone coordinate system.

Images