Background technology
The research and development of current geometric sense precision measurement, from range ability angle, is just experiencing the evolution from stock size to small scale size and large space size both direction.Large space size is far longer than general maching dimension, and conventional precision measurement method can not solve the measurement problem in engineer applied very well; Larger workpiece (product) because size and quality huge, surveying work is many in processing and manufacturing site environment, even carries out on station, complicated space, in addition the uncontrollable on-the-spot disturbing factor such as temperature, vibration, considerably increases measurement difficulty; In addition, huge measurement space and complicated site environment propose very high requirement to Automatic survey level, and it is also one of key factor of large-scale metrology that high-efficient automatic is measured.
Along with country is to the support energetically of the major projects such as large Aircraft Project, wind power generation and deep-sea detecting and development, increasing parts relate to the complex-curved manufacture of large scale and measuring technique, such as aircraft fanjet blade, helicopter screw propeller, submarine screw propeller and wind-powered electricity generation turbine blade etc.These parts have strict requirement for the precision of last shaping curved surface, the product curve form produced should with the shape strict conformance designed, therefore the crudy for this class part should be carried out measuring and detect to ensure the consistance with design parameter, thus guarantees the performance of product.
Large scale Curved dectection major part in current commercial Application is all utilize the three-dimensional coordinates measurement technology of contact to detect, this method completes the measurement of cross-sectional geometry by the measurement and assessment of effects on surface point coordinate, this measuring method precision is high, but measuring speed is very slow, manual intervention is a lot, simultaneously poor universality, operating process redundant and complicated.And this measuring method is generally based upon on the basis of known priori cad model, for being a great inconvenience property of reverse-engineering.
Summary of the invention
In order to realize measuring efficient, high precision, the high universalizable of large complicated part, the object of this invention is to provide a kind of large complicated parts measurement method based on Robot Visual Servoing of the full intellectualized precision measurement without the need to priori cad model.
The present invention solves the scheme that its technical matters adopts: the measure field first setting up loading more than 4 robotic arms, area-structure light is adopted to mark piece surface, then adopt vision technique to set up space curved surface stream shape model, then precision zone division is carried out for curved surface stream shape model and gets some planning automatically; Carry out task matching and get a path planning to again each robot, finally adopt the method control terminal probe of power feel-vision fusion servo to carry out contact and get a little, to obtain the cloud data of large complicated part.
Described space curved surface stream shape model can adopt B-spline control vertex to mate in binocular image and build accurate model to the geometric configuration of tested curved surface, adopts the mesh shape of structured light projection measure the Curvature varying in each region of curved surface and set up the space manifold of curved surface.
Described precision zone divides and is situation according to Curvature varying on space curved surface and carries out Region dividing to curved surface, namely change according to the curvature of curve of grid, and the feature angle point comprising structured light node and curved surface forms grid node density, thus curved surface is divided into the precision zone gone out by feature points segmentation.The generating laser producing area-structure light becomes 25-75 degree to piece surface surface of emission structured light with surface level, structured light plane and piece surface intersect straight line or curve.Described generating laser both sides arrange video camera by vertical direction is equidistant, to launch equidistant parallel surface structured light.
Measure field of the present invention can comprise four removable guide rails and more than 4 servo robot, and servo robot and global measuring control to locate by indoor iGPS.Described measure field also can comprise four fixed guides and four removable guide rail Special composition structural framings, wherein two guide rails can move by vertical direction, two other guide rail can tangential movement, each removable guide rail assembles two mobile robots, overall camera system and indoor iGPS position system device are housed above measurement space.Carry out task matching to each robot by precision zone, each robot completes return measurement data behind subtask.Robot is in the process of executing the task, the shortest in task matching principle using robot displacement, and circulation optimum solution is as optimal allocation method.Robot progressively selects to carry out measurement task with regard near field according to task matching, each robot is measured in the region that chosen distance regional center is minimum in the course of the work, the region that after having measured, chosen distance is minimum again, so circulation is until all area measures complete.
Another object of the present invention is to provide a kind of special purpose device for the large complicated parts measurement method based on Robot Visual Servoing, it comprise support 9, can up and down or the guide rail moved left and right, guide rail is provided with more than 1 vision robot 8, be separately installed with area-structure light in support 9 side and launch battle array A2 and area-structure light transmitting battle array B3, and at the top of support 9 or arranged outside overall situation video camera 1.Guide rail moving up and down comprise be arranged on support 9 two ends move up and down guide rail 4, the guide rail that can move left and right comprise be arranged on support 9 both sides move horizontally guide rail 5.Move horizontally guide rail 5 two ends can respectively with move up and down the activity of 4 one-tenth, guide rail and be connected.
The invention has the beneficial effects as follows: this method be a kind of without the need to part C AD prior model, intelligent planning path, realize high-precision complex parts measuring method, because this method sets up the measure field of oneself, and be mounted with iGPS position measuring station, overall video camera and multiple servo robot, can carry out planning from the overall situation and carry out accurate operation from local, realizing efficient, high precision, high universalizable measurement.The key components and parts that this method can be widely used in China's Major Strategic equipment manufactures field.
Embodiment
In order to illustrate technical scheme of the present invention and technical purpose, below in conjunction with accompanying drawing and specific implementation method, the present invention is described further.
Embodiment 1, first the present invention will set up the measure field (the present invention is described for 8) of loading more than 4 robotic arms, area-structure light is adopted to mark piece surface, then adopt vision technique to set up space curved surface stream shape model, then precision zone division is carried out for curved surface stream shape model and gets some planning automatically; Carry out task matching and get a path planning to again each robot, finally adopt the vision power that power feel-vision merges contact to feel that Hybrid mode measuring method carries out large complicated part measuring the related data obtaining large complicated part.Consult Fig. 1 to Figure 10.
Embodiment 2, space curved surface stream shape model of the present invention can adopt B-spline control vertex to mate in binocular image builds skeleton pattern to the geometric configuration of tested curved surface, adopts the mesh shape of structured light projection measure the Curvature varying in each region of curved surface and set up the space manifold of curved surface.Consult Fig. 1 to Figure 10, all the other are with embodiment 1.
Embodiment 3, precision zone of the present invention divides and is situation according to Curvature varying on space curved surface and carries out Region dividing to curved surface, namely change according to the curvature of curve of grid, and the feature angle point comprising structured light node and curved surface forms grid node, thus curved surface is divided into the precision zone gone out by feature points segmentation.Consult Fig. 1 to Figure 10, all the other same above-described embodiments.
Embodiment 4, the generating laser that the present invention produces area-structure light becomes 25-75 degree to piece surface surface of emission structured light with surface level, structured light plane and piece surface intersect straight line or curve.The equidistant setting of vertical direction pressed by described generating laser, to launch equidistant parallel surface structured light.The overall video camera demarcated obtains part image perpendicular to level ground, and image comprises the intersection of piece surface and area-structure light and piece surface.Visual identification technology is adopted to analyze the image that overall video camera obtains, based on B-spline curves and Snake algorithm, set up the space manifold model of tested curved surface, by the Curvature varying analysis to structured light grid node and mesh lines, judge the degree of crook of piece surface, thus select suitable to get dot density.Feature angle point along stream shape model carries out getting point path planning, according to the difference process of getting dot density and require to carry out varying number, obtains unique point more how to be measured, thus accurately measures.It mainly adopts power feel-vision fusion contact method to carry out getting a measurement.Consult Fig. 1 to Figure 10, all the other same above-described embodiments.
Embodiment 5, measure field of the present invention can comprise four removable guide rails and eight servo robot, and servo robot and global measuring control to locate by indoor iGPS.Consult Fig. 1 to Figure 10, all the other same above-described embodiments.
Embodiment 6, measure field of the present invention also can comprise four fixed guides and four removable guide rail Special composition structural framings, wherein two guide rails can move by vertical direction, two other guide rail can move horizontally, each removable guide rail assembles two mobile robots, for the automatic measurement of follow-up whole system provides hardware foundation.Overall camera system and indoor iGPS position system device are housed above measurement space.Carry out task matching to eight robots by precision zone, each robot completes return measurement data behind subtask.Robot is in the process of executing the task, the shortest in task matching principle using robot displacement, and circulation optimum solution is as optimal allocation method.Robot progressively selects to carry out measurement task with regard near field according to task matching, each robot region that chosen distance regional center is minimum in the course of the work, the region that after having measured, chosen distance is minimum again, and so circulation is until all area measures complete.The data that the present invention is intended to the measurement of each robot obtains merge, and finally obtain the cloud data of piece surface.Consult Fig. 1 to Figure 10, all the other same above-described embodiments.
Embodiment 7, needs to install overall video camera and indoor iGPS device above the measurement space that the present invention proposes, the base of eight robots is equipped with indoor iGPS device, to obtain the accurate location of each robot after moving on rails.
First, tested part is placed into the measure field established, face laser transmitter group is to piece surface surface of emission laser, face laser is crossing with piece surface obtains Laser Curve, in order to the degree of crook enabling the curve of acquisition react piece surface in the image that video camera obtains, face laser transmitter group parallel equidistant is from vertical arrangement, and the face laser launched becomes 25-75 degree angle with surface level; Then by overall camera picked-up part general image, by demarcating the shape and the profile that obtain part, thus set up Snake energy model objective contour curve is approached, profile is considered as an energy minimization batten, adopt B-spline curves to be used for contour detecting and matching fast, thus obtain shape and the profile of part.Adopt and the similar energy function of Snake, the objective contour evolution forms of initial B-spline curves represents, and the external energy term of curve is intended adopting GVF(gradient vector flow) external energy expression represent.The equation concrete according to B-spline curves can derive the Snake model that corresponding B-spline curves approach image outline.In binocular image, mate by B-spline control vertex the shape calculating tested curved surface, measure the Curvature varying situation in each region of curved surface by the mesh shape of structured light projection on this basis and set up the space manifold of curved surface.
According to the profile on space curved surface, and mesh shape and node density determine three accuracy classes, are respectively: superhigh precision, high precision, normal precision.Superhigh precision region refers to that mesh shape distortion is very large, and the region that node density is also very large, illustrates the piece surface of one's respective area be out of shape or degree of crook larger, need to carry out precision and get the shape facility that an ability accurately obtains piece surface; High precision region refers to that mesh shape distortion is comparatively obvious, and the region that node density is also larger, illustrates that the piece surface of one's respective area has distortion or bending existence, needs to measure the shape facility determining piece surface compared with multi-characteristic points; Normal region refers to that distortion of the mesh is not obvious, the region that node density is less, illustrates that the piece surface of one's respective area is closer and can measure less unique point like plane characteristic, improves measurement efficiency to reduce workload.
Measurement space of the present invention carries eight robots altogether, and label is one to No. eight respectively, and wherein one three five No. seven artificial six articulated robots of four machines, carry out the measurement task in superhigh precision and high precision region; Artificial four articulated robots of all the other four machines, carry out the measurement task in normal precision region.After piece surface to be measured is divided into three accuracy classes, guide rail moves to appropriate location according to part height and then maintains static.Be A, B, C by superhigh precision, high precision and normal precision three zone markers respectively, again six articulated robots and four articulated robots are labeled as S and F respectively, by calculating the distance of each robot S to A and B, the shortest one group of coupling of chosen distance is measured, and then minimum one group of chosen distance is measured from remaining Different matching distance, by that analogy.While six articulated robot work, the shortest coupling of distance is carried out in four articulated robot F and C region.If the same guide rail of robot exists in the course of the work the situation of mutually interfering, then first walk off from one's job task, after first round measurement task completes, more than repeating after moving guide rail, measures task matching, until complete the measurement task in all regions, tested part surface.
When robot is on guide rail after mobile end, first will be obtained the position data of pedestal by indoor iGPS device, then fixed pedestal, by the rotation of joint of robot or mobilely carry out surveying work.
Point process is got for probe adopt the method control terminal probe of power feels-vision fusion servo to carry out contact to get a little.Getting point process for making contact type probe contacts soft, power is felt control is incorporated in binocular vision servo-control system, binocular vision and power visual information is combined together.The parallax between Corner and roboting features's point is drawn by two ccd video cameras; Then reduce the value of binocular parallax until zero (namely the unique point of robot and Corner overlap) by the motion of control, thus reach the object of hand eye coordination control.As shown in Figure 6, the angular parallax of two unique points on target object and tail end manipulator of robot is obtained by stereo visual system
with
, the object of Visual Feedback Control makes exactly
with
all reduce to trend towards 0, as shown in Figure 7, namely make two unique points overlap, also namely robot end reaches target location.When probe does not touch curved surface, the feedback quantity f of force snesor is 0, now controls according to binocular vision servo, as shown in Figure 8; When power controls as the outer circulation controlled to reduce the contact force of probe and tested curved surface to during measured surface by probes touch.
Whole measuring process is by overall camera control, indoor iGPS position measuring station gathers the pose data of robot, and the measurement data obtained by each region of computer recording, after the surveying work of regional completes, the data analysis measured also is generated a some cloud, based on the automatic data acquisition end-of-job that Robot Visual Servoing controls.Consult Fig. 1 to Figure 10, all the other same above-described embodiments.
Embodiment 8, special purpose device for the large complicated parts measurement method based on Robot Visual Servoing of the present invention, it comprise support 9, can up and down or the guide rail moved left and right, every root guide rail is provided with more than 1 vision robot 8, also can root task need to arrange 3 or 4 vision robots 8, be separately installed with area-structure light in support 9 side and launch battle array A2 and area-structure light transmitting battle array B3, and at the top of support 9 or arranged outside overall situation video camera 1.Guide rail moving up and down comprise be arranged on support 9 two ends move up and down guide rail 4, the guide rail that can move left and right comprise be arranged on support 9 both sides move horizontally guide rail 5.Move horizontally guide rail 5 two ends can respectively with move up and down the activity of 4 one-tenth, guide rail and be connected.Consult Fig. 2 to 5.All the other same above-described embodiments.