Patent Application
20100145509
This invention relates to robotic assembly and more particularly to in-production optimization of the parameters associated with such assembly.
Industrial robots with force control are used more and more in assembly applications in both the automotive and general industries. These assembly applications involve the robot assembling the same set of parts in continuous product runs. For example and without limitation, the robot may be assembling the parts used in an automobile transmission or other powertrain components.
U.S. Pat. No. 6,553,652 and U.S. Pat. No. 7,181,314 describe the use of an industrial robot with force control to perform tight-tolerance assembly tasks that cannot be performed by a conventional industrial robot with position control. However, since the introduction of force control, in which the actual robot path depends not only on the programmed position but also on the interaction force between the assembled parts/components (hereinafter “parts”), the optimal set of robotic (force control) parameters is often obtained either by trial and error or offline analysis tools. This parameter selection process is tedious and time consuming.
The Patent Cooperation Treaty (“PCT”) application published as WO2008/085937 (“the PCT Application”), discloses a method and system for robotic assembly parameter optimization. As disclosed therein, a force controlled robotic assembly process is first categorized into different assembly types such as cylindrical, radial and multi-staged. Then each type of assembly is further parameterized into robotic assembly parameters such as Search Force, Rotation Angle, Rotation speed, Force Amplitude, Force Period, and so on.
In the robotic assembly parameter optimization methods and systems described above, the optimization process needs to be performed prior to the use of the robot in the assembly process. The optimization experiments need to be performed off-line and a limited number of parts are used in the experiments. Therefore the optimization results can be close to but not exactly correct for the production environment since there are manufacturing variations and differences in the assembly cell setup. Furthermore, some production related process parameters such as assembly starting position cannot be obtained prior to production. Thus in-production assembly parameter optimization is desired. The present invention provides an in-production or, as it is sometimes called, an on-line robotic assembly parameter optimization.
A computer program on a tangible media for optimizing parameters of a robot. The robot parameters are associated with repeated assembly of parts by the robot during a continuous production run of the parts to be assembled. The robot uses the robot parameters to be optimized to repeatedly assemble the parts during the continuous production run of the parts to be assembled.
The computer program on the tangible media has:
computer usable code configured to simultaneously optimize the robot parameters to be optimized during the repeated assembly of the parts by the robot.
A system for repeatedly assembling parts during a continuous production run of parts to be assembled. The system has:
a robot for assembling the parts. The robot has parameters associated with the parts assembly. The robot uses the parameters to repeatedly assemble the parts during the continuous production run of the parts to be assembled; and
a computing device connected to the robot to control the robot during the repeated assembly by the robot of the parts, the computing device having a computer program therein having program code configured to simultaneously optimize the robot parameters associated with the parts assembly while the robot repeatedly assembles under control of the computing device the parts using the robot parameters associated with the parts assembly.
In a system for using a robot to repeatedly assemble parts during a continuous production run of parts to be assembled, the robot having parameters associated with the parts assembly, a method for optimizing the parameters.
In the method:
the robot uses the robot parameters to be optimized to repeatedly assemble the parts during the continuous production run of the parts to be assembled; and
simultaneously optimizing the robot parameters to be optimized during the repeated assembly of the parts by the robot.
a) shows the relationship between the two parts to be assembled by a robot.
b) shows the two parts after they are assembled.
c) shows the variation of the relative position of the parts to be assembled prior to their assembly.
d) shows the distribution of the X locations of that one of the two parts to be assembled whose position can vary prior to assembly.
Referring now to
Four subsections are included herein to describe the present invention. These subsections are: 1) the method and system to setup and switch between production and parameter optimization runs; 2) the method to select the parameter boundaries and constraints including robot search parameters, assembly termination condition and assembly starting position; 3) the method to optimize the parameters; and 4) the method to optimize general process parameters such as starting position. All of these methods are performed either on the robot teach pendent 101 or a PC, not shown in
1) The method and System to Setup and Switch Between Production and Parameter Optimization Runs
The setup and switching method and system includes the graphical user interface on the robot teach pendant 101 or a PC connected to controller 102, the main page 200 of which is shown in
When the user requests the execution of an in-production parameter optimization/learning task, for example, by checking the box for Starting Position Learn 202, a predefined number of learning runs are conducted with the production parameter set and the positional information is recorded and stored. The criteria for setting the number of predefined learning runs is the random deviation between the taught start position of the assembling or as is it also known as the inserting part and the coming-in part, that is the assembled or as it is sometime known as the inserted part, location at the assembly station on the production line and the desired accuracy of the learning. The assembled parts are coming in continuously from the production line.
As is well known, the higher the number of learning runs the better is the accuracy of the results. What is learned during the learning runs are the starting position parameters associated with the real world parts to be assembled. One example of a technique for learning the starting position parameters from the assembly end position is described below in subsection 4.
The data is analyzed after the learning runs are completed in teach pendant 101 or a PC connected to controller 102 using the method described below. The result of the analysis is presented graphically on a display device such as the teach pendant 101.
When the parameter optimization box 204 is checked, the assembly parameter values are varied in a predefined manner during the optimization runs and the values and the resulting cycle time and/or success rate may be stored in a data file for later processing. The predefined manner for varying the assembly parameter values, that is, the optimization method, may, for example, be the well known Design of Experiments (DOE) method described in detailed in the PCT Application. The user of the present invention has the ability to vary the selected optimization method or change to another optimization method through the setup interface. The parameter optimization is performed in parallel with the production operation. The parameter optimization can be set to be performed automatically and thus performed continuously or as is described below with reference to the flow diagram of
There are a predefined number of optimization runs which are determined by the predefined manner and are based on the number of parameters to be optimized and the levels of the parameter values that are varied within a certain range to find the values that are the best in a statistical sense. After the predefined number of runs are completed, the system automatically returns to the production parameter set and continues the normal production operation. At the same time, the data arising from the optimization runs is processed on a CPU in the teach pendant 101 or another computational device such as a PC connected to controller 102 using the method described in a later section of this Detailed Description.
When the box for Parameter Verification 206 is checked, the verification process is executed. The parameter verification process uses the starting position and/or parameter set obtained from the position learning and parameter optimization. A relatively large number of repeated experiments for the parameter verification process, such as for example 100, are necessary to obtain results that are statistically reasonable. The data processing and presenting methods are described below.
The parameter verification process of the method and system of the present invention determines if the optimized parameters and learned starting positions are to be used for the next round of production runs. The indicators used in making the determination are the assembly cycle time and the change in the assembly success rate threshold. A cycle time that is lower is better than a higher cycle time while the assembly success rate that is higher is better than a lower success rate. The current and previous parameter/position data are stored in a file. A supervision option is also provided to allow the operator to determine if the optimized parameters are used in the next round of optimization process. This option is shown in detail in
When the learning runs are finished, that is, verified, the system returns to its normal production run and asks in query (309) if the starting position is to be 10 updated to the new value. If the answer is Yes, the starting position is updated with the learned data and the process proceeds to the parameter optimization query 304. If the answer to 309 is No, the starting position remains unchanged and the system proceeds directly to the Parameter Optimization query 304.
If the answer to the parameter optimization query 304 is Yes, the parameter optimization is performed in 305. When the parameter optimization runs are finished the system returns to normal production runs and uses the optimized parameters only if the switch of the parameter Verify query 306 is No and the switch of the Update Param query 308 is Yes. Otherwise the production parameters are used after the optimization runs. The optimization data is processed and the result is stored in a file. If the answer to the parameter optimization question 304 is No, the parameter optimization is skipped.
When the answer to Parameter Verification 306 is Yes, the verification runs 307 are conducted. At the end of the verification runs, the system is switched back to the normal production run. The verification data is stored and processed and the process proceeds to the next question 308 which is should the production parameters be updated? If the answer to 306 is No, the verification process 307 is skipped and the process proceeds directly from query 306 to query 308. If the answer to the updating the parameter query 308 is Yes, the production parameters are updated to the optimized parameters. If the answer to 308 is No, the production run continues with the existing parameters. The optimization/learning process can be repeated as the batch and other production environments change. As was described above, the method used in verifying the parameter optimization result is also the same method that is used for verifying the learning result.
As is shown in
It should be appreciated that in the present invention as is shown by the flow diagram of
It should also be appreciated that the technique of the present invention is running during the production and the user can direct the process flow by use of the switches. When all the question switches 302, 304, 306308 and 309 are set to No or Hold, the system runs with the default production parameters.
One of the major differences between the off-line assembly parameter optimization that is disclosed in the PCT Application and the present invention of in-production assembly parameter optimization is that in the present invention parameter boundaries and other constraints are more important than in the prior art off-line optimization. Therefore, the variation of the assembly parameters should be controlled to minimize the impact to the normal production throughput and assembly quality.
However, if the variation of the assembly parameters is too small, they may not be distinguished from the manufacturing system noises. Thus the optimized result may not reflect the nature of the real system. If the variation of those parameters is too large, the normal production could be interrupted to a considerable degree, causing an unacceptable failure rate and/or a long cycle time.
Constraints are also needed to be applied to some of the variables such as for example maximum searching force, the minimum value of the robot force control damping factor. Those constraints are necessary for the system to have the assembly quality fall within predetermined boundaries. These parameter boundaries and variable constraints can only be obtained from the real-world manufacturing process.
Real-case testing 403 is conducted and the resultant data is collected and recorded. The data is then fed into an estimator 404 to obtain a new set of parameter boundaries and variable constraints. The estimator 404 determines if the data meets the criteria described in the paragraph above and if not the estimator 404 provides a new set of parameter boundaries and variable constraints that can meet that criteria. This process can be repeated until a set of process dependent parameter boundaries and constraints are obtained. The user can overrule the estimator's parameter boundary values by using the graphical user interface 401.
Any optimization technique can be used in this in-production parameter optimization. The well known DOE (Design Of Experiments) method, described in detail in the PCT Application, is an example of one such optimization technique.
In addition to the basic robotic force control assembly parameters (for searching and insertion), there are other assembly process parameters such as starting position that can be optimized (improved or learned) during production. Starting position is used as an example in
a) shows the relationship between the inserting part 501 and the inserted part 502; 5(b) shows that relationship when the parts 501 and 502 are assembled; 5(c) illustrates the variation of the relative positions of the parts 501 and 502 at the beginning of the assembly while the starting position of the inserting part 501 is fixed (taught) but the position of the inserted part 502 is varied since the pallet (not shown in
Using a statistical method, the actual mean and deviation of the X-Y position at the end of insertion for the inserted part 502 can be calculated. This data is used to modify the starting position of the inserting part 501 to get a better estimation of the “right” X-Y positions for a quick and a statistically higher success rate and shorter assembly cycle time.
d) illustrates the distribution of the X locations of the inserted part 502. The plus signs 503 illustrate the recorded end positions of part 502 in a number of different assembly operations and the curve 504 shows the distribution that is derived from those end positions for the X-coordinate. The assembly start position is taught and relatively fixed in space. That position is indicated by the solid line 508. The distance between the X-coordinate of the starting position 508 and the X-coordinate of the average end position as indicated by the dashed line 507 is Dx 509. Thus the X coordinate of the starting position of inserted part 502 can be shifted to the left by Dx to obtain a more accurate assembly starting position. For positions in the Y-coordinate, a similar analysis can be applied to derive the shifting distance Dy (not shown in
Referring now to
As can be appreciated by one of ordinary skill in the art, the present invention may take the form of a computer program product on a computer-usable or computer-readable tangible medium having computer-usable program code embodied in the medium. The computer-usable or computer-readable medium may be any tangible medium that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device and may by way of example but without limitation, be an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, device, or even be paper or other suitable medium upon which the program is printed. More specific examples (a non-exhaustive list) of the computer-readable medium would include: a portable computer diskette, a flash drive, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, or a magnetic storage device.
Computer program code for carrying out operations of the present invention may be written in an object oriented programming language such as Java, Smalltalk, C++ or the like, or may also be written in conventional procedural programming languages, such as the “C” programming language, and robot program language such as “RAPID” —an ABB proprietary robot program language. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider).
It is to be understood that the description of the foregoing exemplary embodiment(s) is (are) intended to be only illustrative, rather than exhaustive, of the present invention. Those of ordinary skill will be able to make certain additions, deletions, and/or modifications to the embodiment(s) of the disclosed subject matter without departing from the spirit of the invention or its scope, as defined by the appended claims.
