Mobile augmented reality for mobile robot assembly manual

Abstract

In industrial operations such as assembly tasks, manuals or guidebooks are commonly used to convey information via visual methods in the form of text, pictures or videos and more recently Augmented Reality (AR) which has gained traction in recent years. However, several researchers highlighted their studies may not accurately represent actual assembly scenarios in terms of complexity as they use toys such as LEGO or puzzles or short assembly durations which may not show the effects of AR on the participant for longer assembly processes. The objective of this research is to design a developers-based algorithmic conversion of content in paper manual to AR-based manual, to develop a modified paper manual and AR-based manual for mobile robot assembly, and to evaluate the performance of paper manual, modified paper manual, and AR-based manual in mobile robot assembly. In this research, the mobile robot Turtlebot was used as the testbench which is a widely used mobile robot in research and requires approximately 3 hours for assembly. To develop the content for AR-based manual, a developers-based algorithmic technique from literature was adopted for textual instructions and modified to extract and convert visual information from paper manual. A study was then conducted to evaluate the effectiveness of three types of manuals which are paper manual, modified paper manual, and AR-based manual on participants’ assembly performance. Overall, participants using AR-based manual completed the assembly faster by 21.72% while modified paper manual have an improvement of 7.5% compared to paper manual. In terms of number of assembly errors, AR-based manual and modified paper manual were found to have similar performance where the number of assembly errors is at 2.25 and 2 respectively while average assembly errors for paper manual is 5. NASA- TLX scoring showed higher mental workloads but lower frustration and effort from AR-based manual participants compared to other manuals. User satisfaction score is also highest for AR-based manual participants from the Likert-scale questionnaire. The contribution of this study is on the introduction of buffer steps during conversion of paper to AR-based manual which can be adopted for other paper manuals and the evaluation of participants’ performance using longer and more complex assembly tasks. AR-based manual participants were also concluded to have an overall better performance.