Production Scenario
According to the User Centred Design approach the project is following, early field studies with real end-users were designed. This first prototype formed the basis to evaluate the different modalities of interaction with the assembly line workers under real conditions in the plant Skoda has in Vrchlabi, Czech Republic. The main objective was to obtain the users’ feedback, regarding their preferences and attitudes towards several hardware, software, user interface and remote support-related-features. The chosen mechanisms for HCI were voice recognition, textile keyboard, non-explicit or task recognition-based interaction, and Head Mounted Displays. The results obtained had to be useful for future analysis on the impact of wearable technologies in the training process itself.
The front headlight assembly process was selected as a test case, as it represents a complex enough task which justifies the use of wearable technologies during training. After a detailed analysis, the process was broken down into 14 elementary steps that workers had to follow, using several tools, fixtures, accessories and measuring gauges. The focus of the test case was centred on aspects related to the interaction between the worker and the environ-ment and on task detection. A simple software prototype was created that allowed the navigation through the assembly procedure.
Due to the fact that some of the functionalities were not available for the time the study was carried out, the technique known as Wizard of OZ was applied to simulate those features. During the tests, the workers wore as a wearable device a Xybernaut V computer unit (where the application is installed), a Microoptical SV6 Head Mounted Display, and a conventional headset. Besides, there was also a VNC application running on the Xybernaut. This allowed a remote control and interaction with the wearable application from an external computer by the Wizard. All 14 assembly procedure steps were documented using text, video and photos. The users had the freedom of selecting the information format they found most comfortable, while any kind of combination was allowed, such as text + pictures, text only or video only. All supporting material was structured using XML. Finally, the whole training process was recorded with some users learning in the traditional way, namely without wearable technology. After that, a new group of workers performed the whole training process using the prototype. The test runs were initiated with a short presentation and a pre-questionnaire dealing with ethnographic aspects, work experience and technical skills.
During the voice recognition based interaction, the workers navigated through the application using voice commands. Since no Czech recognition system was available at the time, the wizard interacted remotely with the application according to the workers’ commands. Afterwards, the workers used a textile keyboard for interaction. As there was also no ‘real’ keyboard available, a set of keyboards, adhered to different parts of the workers’ body, were constructed out of paper.
- The textile keyboard mock-up
The wizard observed the worker and interacted with the application remotely using the VNC client. Finally, non-explicit interaction was tested. Although there was a first prototype for context detection available, it was not suitable for usability testing. Each time the worker finished one action, the wizard pushed the ‘next’ button remotely in order to present the information corresponding to the next step. In case of error, the wizard simulated the error detection and a message was shown on the screen. The duration time and number of errors during the assembly was recorded. Once finished the test, the workers were asked to complete a usability questionnaire.
- The application used during the first demonstrator
