Work Instructions

The assembly of components in the aerospace industry is currently supported through procedures based on the generation of work instructions. This documentation describes both the sequence of operations to be performed by operators, and fundamental and critical parameters of operation (drawings of components, torques to be applied, sealing system characteristics or paste, etc.). Currently, workers use this information to ensure that the tasks are performed correctly. However, sometimes the documentation is difficult to manage, either by the difficulty of interpreting the drawings or because the process is too complex, for example in the installation of electrical harnesses. This document shows the results of the Project MOON (asseMbly Oriented authOring augmeNted reality) developed by AIRBUS Military in 2010. MOON uses 3D information from the iDMU (industrial Digital Mock-Up) to generate assembly instructions by applying Augmented Reality technology. A demonstrator was developed for the electrical harness routing in the frame 36 of the AIRBUS A400M. The techniques and methods here described are 'patent pending'.

It is widely accepted that presenting data in the form of pictures or models can enhance comprehension, decision making and communication of the underlying information. However, there are few systematic studies that examine whether graphical models are more effective than other representation (such as textual descriptions). Process models provide an abstract graphical view of organizational procedures by reducing the complex reality of the work performed by a company to its most important activities. Such models are useful to train new employees, to document and allow improving organizational procedures and policies. This paper describes an experiment to address if there are significant differences in terms of process understand ability depending on whether textual work instructions or process models are used to represent a business process. We compared a control group of subjects that received textual work instructions to a second group, which received process models, in terms of their ability to understand the process. We found empirical support that using textual work instructions or process models do not influence process understand ability for non-expert users but do influence for experienced users.

The progressive development of human-robotcollaboration (HRC) during the last decade, and latest in the context of the fourth industrial revolution (Industrie 4.0), opens up a wide range of possibilities for integrating people with disabilities and elderly people in complex production processes. In order to support these user groups by compensating individual limitations, it is necessary to develop new multimodal interaction strategies as well as task allocation and orchestration concepts with special attention to ergonomics, personalization, and adaptability. Our research targets on transforming general work instructions of a HRC work process to individual and useroriented instructions and presenting them using accessible multimodal user interfaces, such that users with disabilities can participate in the process and, in cooperation with the robot, perform value-adding production processes. This publication describes a system approach to meet these requirements, illustrates a possible architecture and validates its suitability for use by presenting an implementation of a proof-of-concept prototype.

Increased product complexity and the focus on zero defects, especially when manufacturing complex engineered products, means new tools are required for helping workers conduct challenging assembly tasks. Augmented reality (AR) has shown considerable promise in delivering work instructions over traditional methods. Many proof-of-concept systems have demonstrated the feasibility of AR but little work has been devoted to understanding how users perceive different AR work instruction interface elements. This paper presents a between-subjects study looking at how interface elements for object depth placement in a scene impact a user’s ability to quickly and accurately assemble a mock aircraft wing in a standard work cell. For object depth placement, modes with varying degrees of 3D modeled occlusion were tested, including a control group with no occlusion, virtual occlusion, and occlusion by contours. Results for total assembly time and total errors indicated no statistically significant difference between interfaces, leading the authors to conclude a floor has been reached for optimizing the current assembly when using AR for work instruction delivery. However, looking at a handful of highly error prone steps showed the impact different types of occlusion have on helping users correctly complete an assembly task. The results of the study provide insight into how to construct an interface for delivering AR work instructions using occlusion. Based on these results, the authors recommend customizing the occlusion method based on the features of the required assembly task. The authors also identified a floor effect for the steps of the assembly process, which involved picking the necessary parts from tables and bins. The authors recommend using vibrant outlines and large textual cues (e.g., numbers on parts bins) as interface elements to guide users during these types of “picking” steps. Disciplines Electrical and Computer Engineering | Ergonomics | Mechanical Engineering | Operations Research, Systems Engineering and Industrial Engineering Comments This proceeding is published as MacAllister, Anastacia, Melynda Hoover, Stephen Gilbert, James Oliver, Rafael Radkowski, Timothy Garrett, Joseph Holub, Eliot Winer, Scott Terry, and Paul Davies. "Comparing Visual Assembly Aids for Augmented Reality Work Instructions." In Proceedings of the 2017 Interservice/ Industry Training, Simulation, and Education Conference (I/ITSEC). Volume 2017, Paper no. 17208. Arlington, VA: National Training and Simulation Association. Posted with permission. Authors Anastacia MacAllister, Melynda Hoover, Stephen Gilbert, James Oliver, Rafael Radkowski, Timothy Garrett, Joseph Holub, Eliot Winer, Scott Terry, and Paul Davies This presentation is available at Iowa State University Digital Repository: https://lib.dr.iastate.edu/me_conf/197 Interservice/Industry Training, Simulation, and Education Conference (I/ITSEC) 2017 2017 Paper No. 17208 Page 1 of 14 Comparing Visual Assembly Aids for Augmented Reality Work Instructions Anastacia MacAllister, Melynda Hoover, Stephen Gibert, James Oliver, Rafael Radkowski, Timothy Garrett, Joseph Holub, Eliot Winer Scott Terry, Paul Davies Iowa State University The Boeing Company Ames, Iowa St. Louis, Missouri [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], ewin[email protected] [email protected] [email protected]

Augmented Reality (AR) as an assistive technology is a promising tool for novice operators to learn assembly processes. This experiment compared an AR instruction method to display based electronic working instructions (EWI) for product assembly, to assess learning during the first repetitions of the products. In addition, two types of work instructions were used, i.e., standard and chunk instructions. In this experiment a chunk instruction consists of six assembly steps. Effects of the instruction method and type on the learning phase were evaluated with 24 novice operators building two products i.e.. Operators were then asked to build the same products without instructions in order to assess learned skills and establish effects on the recall phase, also as a result of instruction method and type. Task completion time (TCT), product quality, operator workload and learning curve were measured. The learning curve, as indicated by the TCT, took place during the first three repetitions of product assembly. Instruction method and instruction type had no effect on the learning curve. Product quality was high and no differences were found between learning conditions. Operator workload revealed that chunking of the instruction increased workload during the learning phase. During the recall phase, the AR group's TCT increased 19.2%, but only for the first product's repetition without instruction. Product quality remained the same during the recall phase, however operator workload was reduced for chunk learned products. This study indicates that chunking of instructions should be avoided for novice workers. Both EWI and AR can be used for teaching new assembly procedures. While AR and EWI are useful during the learning phase, there are indications that these methods might hinder the operator once they required the necessary skills and knowledge to assemble the product. A possible solution is making instructions more adaptive to fit the skill proficiency of the operator.

A study made by Johansson, Fast-Berglund and Moestam (in press) shows that diversity regarding how information is used exists in global production networks. To be closer to markets, organizations have chosen to globalize their business which is one reason for why diversity arises. This because product types and brands historically have been different. One company that is currently working with improving consistency among processes is Volvo Group Trucks Operations (GTO). The company wants to evaluate how the process for preparing assembly work instructions looks like at different sites within their production network. This enables Volvo GTO to start their work towards a standardized process and uniformity.A starting point for this is to make a current state analysis of the process for preparing assembly work instructions when producing Volvo trucks, engines and transmissions in Sweden. The purpose is to identify key activities within the process and important factors to consider when standardizing the process. This is done on three sites, one for each area.Volvo GTO is the part of the Volvo Group that covers all production of engines and transmissions as well as the production of Volvo, Renault, Mack, and UD trucks. In 2012 the group choose to reorganize from brand based where each brand was an own organization to joint units, for example center of development, operations etc. This has led to a greater need of one common process for preparing assembly work instructions in order to create uniformity among the brands Volvo, Renault, Mack, and UD trucks.The result of the current state analysis shows that the process for preparing assembly work instructions is differently performed depending on the site studied. Despite this, some activities in each process are similar. These were found to be: design, review, time setting, time analysis, balancing, station marking, create assembly work instructions, and share information. Since some activities actually are similar, it would be possible to standardize the process for preparing assembly work instructions in the future. Important to consider when standardizing a process is to create awareness and involvement among employees. It is also important to have the management committed as well as uniformity among IT systems used when performing a process. One last thing to consider is that the process needs to be adaptable because sites are located all over the world and have different culture and regulations.

The objective of this research is to investigate the requirements and performance of parts-of-speech tagging of assembly work instructions. Natural Language Processing of assembly work instructions is required to perform data mining with the objective of knowledge reuse. Assembly work instructions are key process engineering elements that allow for predictable assembly quality of products and predictable assembly lead times. Authoring of assembly work instructions is a subjective process. It has been observed that most assembly work instructions are not grammatically complete sentences. It is hypothesized that this can lead to false parts-of-speech tagging (by Natural Language Processing tools). To test this hypothesis, two parts-of-speech taggers are used to tag 500 assembly work instructions (obtained from the automotive industry). The first parts-of-speech tagger is obtained from Natural Language Processing Toolkit (nltk.org) and the second parts-of-speech tagger is obtained from Stanford Natural Language Processing Group (nlp.stanford.edu). For each of these taggers, two experiments are conducted. In the first experiment, the assembly work instructions are input to the each tagger in raw form. In the second experiment, the assembly work instructions are preprocessed to make them grammatically complete, and then input to the tagger. It is found that the Stanford Natural Language Processing tagger with the preprocessed assembly work instructions produced the least number of false parts-of-speech tags.Copyright © 2015 by ASME

This master‟s thesis is based on a method including interviews and a workshop performed at Autoliv Sweden AB ́s production facility in Vårgårda. The method also includes a contrast study carried out at Scania, IAC and Saab, which all are distinguished companies in the field of standardized work instructions. Autoliv Sweden AB is a manufacturing company that produces safety systems for automotive manufactures around the world. In the production facility, standardized work instructions have been implemented. However, they are not put to correct use, resulting in a production output that varies. Operators perform work tasks differently and thereby do not follow standardized work instructions. The first reason for this is the nonfunctioning system when educating operators in new work task procedures. The second reason why standardized work instructions are not fully used is due to the limited follow-up concerning if operators work according to standardized work instructions. The third reason is because operators are not involved in the development of standardized work instructions which result in low motivation to work according to them. Autoliv Sweden AB wants to use a visualization system to facilitate for operators to practice their work tasks in a correct procedure at work stations. The use of a visualization system is discussed and evaluated in this master‟s thesis as well as how to prevent the three reasons for not using standardized work instructions. This has resulted in this master‟s thesis recommendations which will provide Autoliv Sweden AB with means to ensure that operators work according to standardized work instructions both today and in the future. The recommendations developed in this master‟s thesis result in a training system for operators that are taught new work task procedures. They result in a system on how to perform follow-up regarding if standardized work instructions are used. They result in a new method on how to visualize standardized work instructions as well as a system that includes operators in the development of the standardized work instructions.

Abstract The process of robotics in the world, but also in Slovakia is constantly increasing, and although new types of robots are being developed, the human hand of those work stations cannot be removed completely. Despite the already developed and functioning collaborative robots that allow safe operations side by side with humans, it is necessary for a worker to known accurate, fast and safe procedure of his working activities. Visual work instructions (VWI) are made for this purpose. The worker must be able to quickly understand the principle of work procedure and get quick information about potential risks in very short time. Therefore, great regard for the content of VWI must be taken. At first glance, a simple visual work instructions enable not only to increase safety at the workplace, but also increase workers productivity.

The state-of-practice method to produce formal and good work instructions is time intensive, error prone and produces instructions with inconsistent format and content. These problems force contractors to rely on verbal communication to deliver work instructions to their laborers despite the negative impacts of this informal communication on productivity, rework, workface questions, and safety. This paper presents an automated system – FIPAPM, Field Instructions from Product And Process Models – to produce work instructions for laborers of cast-in-place (CIP) concrete construction. The FIPAPM system extracts design information from product models and construction information from process models based on the work scope of the activity for which the work instruction is produced. The presented system proved to produce work instructions faster, more correctly, and more consistently than the state-of-practice method. Therefore, FIPAPM enables contractors to produce formal, good-quality work instructions on a daily basis reducing field communication problems of informal, verbal work instructions.