外文翻译--使用语音识别技术控制的焊接机器人工作单元

1、精选优质文档-倾情为你奉上Use of Voice Recognition for Control of a Robotic Welding WorkcellABSTRACT: This paper describes work underway to evaluate the effectiveness of voice recognition systems as an element in the control of a robotic welding workcell. Factors being considered for control include program edit

2、or access security,Preoperation checklist requirements, welding process variable control,and robot manipulator motion overrides. In the latter two categories, manual vocal control is being compared against manual tactile control and fully automatic control in terms of speed of response, accuracy, st

3、ability, reliability. And safety.Introduction Voice recognition technology is now recognized as a potential means for easing theworkload of operators of complex systems. Numerous applications have already been implemented, are in various stages of development, or are under consideration. These inclu

4、de data entry,control of aircraft systems, and voice identification and verification for security purposes.Voice control has also been proposed for use aboard the space station. One prime area for application would be control of some functions of robots used for intraand extravehicular inspection, a

5、ssembly, repair,satellite retrieval, and satellite maintenance when a crewmember is serving in a supervisory capacity or the system is operating in a teleoperation mode. Voice control of sensors and process variables would free the crewmembers hands for other tasks, such as direct control or overrid

6、e of the manipulator motion. Similarly, the workload associated with control of many onboard experiments could be eased through the use of this technology.This paper describes the application of voice recognition for control of a robotic welding workcell. This is a complex system involving inputs fr

7、om multiple sensors and control of a wide variety of robot manipulator motions and process variables. While many functions are automated, a human operator serves in a supervisory capacity, ready to override functions when necessary. In the present investigation, a commercially available voice recogn

8、ition system is being integrated with a robotic welding workcell at NASA Marshall Space Flight Center, which is used as a test bed for evaluation and development of advanced technologies for use in fabrication of the Space Shuttle Main Engine. In the system under development, some functions do not y

9、et have automatic closedloop control, thus requiring continuous monitoring and real-time adjustment by the human operator. Presently, these ovemdes are input to the system through tactile commands (;.e. pushing buttons. turning knobs for potentiometers, or adjusting mechanical devices). Since the op

10、erator monitors the process primarily visually, he must either look away from the process to find the proper button or knob or rely on“muscular memory”much as a touch-typist does. In the first case, the time of response to a deviant condition may be excessive. In the second case, there is an increas

11、ed probability of a secondary error being introduced by the operator.A voice recognition system could reduce the response time required from the operator.The probability of pushing the wrong button should similarly be reduced. Also, operator fatigue should be minimized.Since the operator can continu

12、ously monitor the process during override input, the effect of the change can be observed more quickly. Thus, if the desired value is exceeded and reverse correction is required, it should be accomplished more quickly, allowing less overshoot. This reduction in oscillation about the desired value ma

13、kes the system more stable.Another factor that can be improved is operator safety. In a safety-critical situation,the robots operation can be halted immediately by use of the “emergency stop, or E-stop, mode, which is initiated, conventionally, by depressing a large button. If an operator inadverten

14、tly finds himself in a hazardous situation, it may be necessary for him to initiate the E-stop sequence. Should the operator not be within reach of the button,however, he may be unable to take the necessary action, and, as a result, could suffer serious injury. Having the capability of stopping the

15、robot by issuing a voice command could significantly improve the operators safety by enabling him to stop the robot even when not within reach of the E-stop button. Manual corrections are occasionally required to adjust the location at which the weld filler wire enters the weld pool. Proper entry lo

16、cation is absolutely critical to sound weld quality. Adjustments are made either by manually adjusting mechanisms that hold the wirefeed guide tube or by issuing tactile commands to a servomechanism. Use of a voice recognition system could eliminate the need for the operator to place his hand within

17、 the working envelope of the robot end effector or, if servomechanisms are employed,could improve speed of response and stability. Another aspect of robot operation in an industrial environment that is very important is the security of a program editing capability of the system. Under no circumstanc

18、es should any unauthorized person be able to enter this programming mode and alter the robots program. A voice recognition system can provide the necessary security by allowing access only for individuals who are authorized and whose voices can be identified by the system. BackgroundRobotic welding

19、is under development by NASA and Rocketdyne for the automation of welds on the Space Shuttle Main Engine that are presently made manually. The programmability of a robot can reduce the percentage of welding defects through a combination of consistency and repeatability unattainable by its human coun

20、terparts. To do this, the robot is programmed to a nominal weld path and level of weld process parameters (i.e., current, travel speed. voltage,wire addition rate). Some adjustment of these values is often necessary due to conditions changing during the weld. A human making a manual weld accomplishe

21、s this adjustment readily, while a robot must rely on the limited talents of sensors and the ability of the operator to override functions when necessary. System IntegrationThe basic elements of the workcell system are shown diagrammatically in the illustration.The ultimate goal of the system develo

22、pment work in progress is to generate robot manipulator programs and weld process programs off line, download them to the workcell supervisory computer, then use sensor subsystems to make closed-loop corrections to the robot path and process variables. Offline programming is being done with an Inter

23、graph modified VAX 780/785-205 computer system with Interact color graphics workstations. Deviations between the programmed robot path and the actual required path are observed and corrected by a sophisticated vision-based sensor developed for this application by Ohio State University.This sensor sy

24、stem is also designed to permit measurement of the molten weld pool surface dimensions and correct welding current level to maintain the weld pool dimensions within desired limits. Presently, a number of functions are still controlled manually, and manual overrides capability is required for all fun

25、ctions. As stated in the Introduction, use of voice recognition may improve the accuracy and speed of response of these manual overrides. To explore this technology, a Votan VRT 6050 stand-alone voice recognition terminal has been integrated into the workcell. This system provides continuous speech

26、recognition of up to 10 sets of words with 75-150 words per set.The integration of the voice recognition system is broken into analog and discrete signals for control. The voice recognition system connects to the control computer through a standard RS232-C communications link.Discrete Control Signal

27、sIn this project, most of the control circuitry is based on discrete digital signals.This is due to the on/off state nature of the circuits to be controlled in the robot controller.The circuits of the system to be controlled by the voice recognition control computer (VRCC) by discrete signals are th

28、e emergency stop circuit and the positive jog and negative jog circuits for motion control. Since the safety of the operator is paramount in any automated workcell, the voice recognition system should be incorporated as a safety feature. To accomplish this, the VRCC has been interfaced into the work

29、cell emergency stop circuit. The emergency stop circuit in the robotic workcell will shut down the welding process and the mechanical motion of the manipulators. Through the use of a digital signal from the VRCC, a relay is energized that interrupts the necessary circuits in the weld power supply an

30、d robot controller. With the use of the voice recognition system as a safety control for this workcell, we have added a third level of redundancy into the emergency stopping ability of the operator (in addition to the present emergency stop buttons).Manipulator motions are controlled through an axis

31、 select button in conjunction with a positive or negative jog button that is depressed by the operator. Once the operator has selected an axis, he depresses one of the jog buttons for the desired travel distance.This function was selected to be controlled by the VRCC because of its utilization durin

32、g automatic operation of the manipulator to correct trajectory errors. The circuitry necessary to control this operation draws the signal to ground through the activation of relays for the positive or negative jog motion. Because motion is achieved only as long as these signals are active low. they

33、can be controlled by discrete digital signals from the VRCC.Analog Control SignalsThere are many variables that affect the quality of weld during the welding process. but the welding current has the greatest effect over a small range of values. It was for this reason, that the welding current was ch

34、osen to be controlled by the voice recognition system.The welding power supply controls the current level through a voltage circuit that uses a range of 0-10 V DC. These voltage values are converted to current levels from 0 to 300 A for welding. A digital-to-analog converter is used in conjunction w

35、ith a multiplying circuit. The converter allows the VRCC to control a voltage level that is used by the weld power supply to achieve the proper welding current. The multiplier circuit is necessary to allow the weld power supply to be controlled by the other subcontroller used in the workcell.Experim

36、ental InvestigationThe accuracy and speed of response of corrections to robot manipulator motion and welding process variables made with the VRCC are being compared with those made with the original control system. Step input errors to robot motion and welding current are introduced randomly into th

37、e robot program. By graphically recording relevant system output signals,the time required for the operator to detect the change and initiate corrective action may be measured. Response accuracy and stability may also be gaged through similar analysis of the relevant recorded system output signals.C

38、onclusionsFuture work will investigate voice control of welding filler wirefeed speed and location of wire entry into the weld pool. Also to be investigated is voice control of welding arc voltage override. Later, restriction of access to the robot program editor by voice recognition may be implemen

39、ted.The use of voice recognition technology for manual supervisory control of industrial robot systems is very promising. This technology has application for aerospace welding due to the need to have constant human supervision over a multitude of process parameters in real time. Future development o

40、f this technology will permit rapid expansion of its application to both robotic and nonrobotic processes. AcknowledgmentSpecial thanks to Mr. Jeff Hudson of Martin Marietta Corporation for assistance in the preparation of the illustration presented in this article.References 1 C. A . Simpson. hl. E

41、. McCauley. E. F. Rolland. J . C. Ruth. and B. H. Williges. "System Design for Speech Recognition and Generation." Hutnnn Factors. vol. 27. no. 2. pp.115-1-11. 1985.2 National Research Council. Committee on Computerized Speech Recognition Technologies.Automatic Speech Rerop1irior1 in sever

42、e Environments National Research Council.1984.3 E. J. Lerner. "Talking to Your Aircraft." Aerospace America. vol. 24. no. 2. pp. 85-88. 1986.4 J. T. Memlield. "Bosing Explores Voice Recognition for Future Transpon Flight Deck." Ariarinn Week and Space Techno/- og!. vol. 124. no.

43、16. pp. 85-91. 1986. 5 A. Cohen and J. D. Erickson, .Future Uses of Machine Intelligence and Robotics for the Space Station and Implications for the U.S. Economy.'' IEEE J. Robotics and Automarion.vol. SMC-16. pp. 1 11-12 I. Jan.iFeb.19866 "Automation and Robotics for the National Space

44、 Program," California Space Institute Automation and Robotics Panel. Cal Space Repon CS1185-01, Feb. 25, 1985.7 "Advancing Automation and Robotics Technology for the Space Station and for the U.S. Economy." Advanced Technology AdvisoryCommittee. NASA TM 87566. Mar. 1985.使用语音识别技术控制的焊接机

45、器人工作单元摘要：本文论述了使用声音识别技术的焊接机器人工作单元在工作过程中的效果、程序编辑者接近机器人的安全试行运转的必要性焊接过程的控制变量机器人操作者的动作规范等因素给与考虑。在焊接过程控制和操作动作两个方面，按照反应速度定位精确性焊接稳定性焊接可靠性和安全性把人工声音控制与手工触觉控制和完全自动化控制进行了比较。绪论声音识别技术已经成为可能缓解操作者工作负担的一种有潜力的复杂系统。许多应用已经落实,或正陆续开发,或正在研究之中。这些措施包括数据的输入飞机的控制和以安全为目的的语音识别。许多应用语音控制技术还建议用于太空站. 一个主要的应用领域将机器人控制功能用于太空舱内检查、装配、维修

46、、卫星回收、维修卫星,是在船上服务的监督能力和系统运作模式的反馈. 声音感应器和过程控制的变数将使船员影响他手上的其它工作,例如直接控制或推翻的操纵议案。 同样,利用工作量控制机载实验这种技术可以缓解许多工作负担。这份文件描述应用语音识别控制的焊接机器人工作单元。 这是一个复杂的系统,涉及多个传感器及控制投入各种机械操作件和变化多样的工艺参数。虽然许多功能是自动化，且为人类监督管理能力所控制,但在必要时随时准备超越这些功能。 在当前的调查中， 在美国航天局的马歇尔空间飞行中心可供商业使用语音识别系统结合了焊接机器人工作单元的技术，这一技术作为试点的评价和开发先进技术并用于制造航天飞机主发动机。

47、在系统开发中，有些功能尚不具备自动跟踪控制,因此需要不断地人力监测和实时调整操作。目前,该系统投入方案是通过触觉指令(即： 推动按钮. 旋转电位计、或者调整机械装置)。由于操作过程中,主要监测者必须考虑在远离的过程中寻找适当的按钮或把手或靠像打字员一样那种打字时的肌肉记忆。第二种情况，可能由于操作者的的二次反应而增加了错误发生的可能性。一个语音识别系统可减少操作者的反应时间。操作者按错按钮的可能性了同样的也会减少。并且,操作者劳累也会大大减小。由于在方案运行的过程中操作者不断监测，可以更快地观察到运行状况改变所带来的影响。 因此,如果超过了预期值,应该更快纠正,，但不能太过度。 这对减少振荡,

48、使系统更加稳定的实现了预期的价值。另一个因素是可以改善操作者的安全.。在一个安全的紧急情况下，机器人的操作者可以采取紧急停止来停止其运行，这种紧急停止模式一般来说是设置一个大按钮，按惯例是一种经常用的方式。如果操作者无意中发现自己在危险的情况下，这时也许他有必要采取紧急停止这种模式。如果操作者不能够按到的按钮，可他也没有能力采取必要的行动时，这样下去，他可能会受重伤。如果操作者者能通过发出声音指令来停止机器人的运行那将会大大的改善操作者的安全，即使操作者在不能按到紧急停止按钮无法停止机器的情况下也将很安全。手工调整有时候需要适应焊丝填充到焊接溶池中的位置。填充到正确合适的位置是焊接质量的关键。

49、既可通过手工调节机制来控制送丝导管也可给自动控制装置发出移动指令来进行调整。使用语音识别系统可以让操作者者不必再把机器人控制效应得指令文件拿在手中，如自动控制装置被使用，可以改善操作的反应速度和运行稳定性。另一方面，编辑系统程序权限的安全是工业机器人在作业环境中很重要的一个安全。在任何情况下，任何未经授权的人能进入程序编辑模式，并且可以改变机器人的控制程序。 一个语音识别系统，可提供必要的安全，使他们那些久久是获得授权的人的声音，才能被机器人系统识别。背景美国航天局正在开发焊接机器人并且焊接自动化设备来代替目前正在用手工焊接的航天飞机的主发动机。使用该机器人的程序，可以通过用手工来难以做到的焊

50、接一致性和重复操作来达到减少焊接缺陷的比例。为此，焊接机可以编成控制额定的焊接通路和所需要的焊接过程参数，(即焊接电流、焊接速度、焊接电压、送丝速度等)。当焊接条件改变的时候做一些有价值调整是很有必要的。一个人用手工来操作焊接时作出调整是很容易的，但是机器人的调节靠传感器的智能和必要的人工操作者的方案调节。系统综述机器人工作系统的基本情况如图表所示，最终的系统开发工作是编辑操作的程序和焊接过程生产线的控制程序，下载这些程序到控制工作单元的电脑，然后使用子系统传感器修正机器人的运行路径和过程，使其可变。利用VAX 780/785-205电脑连接到彩色图形处理工作站来进行图表处理实现脱机设计。机器

51、人由于程序编辑和实际需要之间的偏差是通过俄亥俄州大学研究的精密的视觉传感器来发现和纠正的。这种传感系统也设计成允许测量焊接溶池表面尺寸和改变电流大小来调节焊接溶池保持理想的形状。目前，仍有许多功能人工控制，而且各个方面的功能都需要人工的操作。如前绪论中所述，引进声音识别技术可以改进人工操作的准确性和反应速度。为研究这项技术，Votan VRT 6050声音识别单机终端被引入到机器人的工作单元中。这个连续的语音识别系统可以提供多达10套，每套有75150句话。把语音识别系统的模拟和离散信号输入控制。语音识别系统通过RS232-C的通信连接到控制主机。图1焊接机器人系统设计离散控制信号在这个项目中

52、，大多数控制电路是基于不同的数字信号。这主要是用在一些国产性质的机器人控制器上的。通过语音识别技术控制的计算机来控制的电路系统是通过一种离散信号来控制，这种信号有紧急停止电路和积极响应和消极响应电路的功能。因为任何自动化工作单元中操作者的安全是必须保障的,所以应把语音识别系统的安全也考虑在内。 为达到这一目标，贞技术已引入紧急停止电路的工作单元。机器人工作单元中的紧急停止电路将会停止焊接过程的终止操作者的操作。通过使用数字贞信号，需要中断焊接动力供电线路和机器人控制器的继电器被广泛使用。由于在这一工作单元中使用的语音识别技术这一安全系统，我们又增加了第三种供选择的紧急停车的方案（除了现在已经有的紧急停车按钮）。方案是通过操作者在轴配合正按钮或负按钮之间选择来实现控制的。一旦操作者选择了轴，它可以在理想的距离之内控制负按钮。这种功能的选用是通过控贞信号来控制的，因为贞信号的使用在自动操作中可以纠正运行的错误。在这一操作中有必要通过继电器的正负极的地面信号来达到目的。只因为这些信号很微弱才能达到目的。他们可以通过贞信号远距离控制。模拟控制信号有很多因素影响焊接过程的质量，但是焊接电流对焊接质量的影响绝不是一个小的因素。正因为如此，所以焊接电流被选择为