Augmented reality at the workplace : a context-aware assistive system using in-situ projection
Markus Funk · OPUS Publication Server of the University of Stuttgart (University of Stuttgart) · 2016
Augmented Reality has been used for providing assistance during manual assembly tasks for more than 20 years. Due to recent improvements in sensor technology, creating context-aware Augmented Reality systems, which can detect interaction accurately, becomes possible. Additionally, the increasing amount of variants of assembled products and being able to manufacture ordered products on demand, leads to an increasing complexity for assembly tasks at industrial assembly workplaces. The resulting need for cognitive support at workplaces and the availability of robust technology enables us to address real problems by using context-aware Augmented Reality to support workers during assembly tasks. In this thesis, we explore how assistive technology can be used for cognitively supporting workers in manufacturing scenarios. By following a user-centered design process, we identify key requirements for assistive systems for both continuously supporting workers and teaching assembly steps to workers. Thereby, we analyzed three different user groups: inexperienced workers, experienced workers, and workers with cognitive impairments. Based on the identified requirements, we design a general concept for providing cognitive assistance at workplaces which can be applied to multiple scenarios. For applying the proposed concept, we present four prototypes using a combination of in-situ projection and cameras for providing feedback to workers and to sense the workers' interaction with the workplace. Two of the prototypes address a manual assembly scenario and two prototypes address an order picking scenario. For the manual assembly scenario, we apply the concept to a single workplace and an assembly cell, which connects three single assembly workplaces to each other. For the order picking scenario, we present a cart-mounted prototype using in-situ projection to display picking information directly onto the warehouse. Further, we present a user-mounted prototype, exploring the design-dimension of equipping the worker with technology rather than equipping the environment. Besides the system contribution of this thesis, we explore the benefits of the created prototypes through studies with inexperienced workers, experienced workers, and cognitively impaired workers. We show that a contour visualization of in-situ feedback is the most suitable for cognitively impaired workers. Further, these contour instructions enable the cognitively impaired workers to perform assembly tasks with a complexity of up to 96 work steps. For inexperienced workers, we show that a combination of haptic and visual error feedback is appropriate to communicate errors that were made during assembly tasks. For creating interactive instructions, we introduce and evaluate a Programming by Demonstration approach. Investigating the long-term use of in-situ instructions at manual assembly workplaces, we show that instructions adapting to the workers' cognitive needs is beneficial, as continuously presenting instructions has a negative impact on the performance of both experienced and inexperienced workers. In the order picking scenario, we show that the cart-mounted in-situ instructions have a great potential as they outperform the paper-baseline. Finally, the user-mounted prototype results in a lower perceived cognitive load. Over the course of the studies, we recognized the need for a standardized way of evaluating Augmented Reality instructions. To address this issue, we propose the General Assembly Task Model, which provides two standardized baseline tasks and a noise-free way of evaluating Augmented Reality instructions for assembly tasks. Further, based on the experience, we gained from applying our assistive system in real-world assembly scenarios, we identify eight guidelines for designing assistive systems for the workplace. In conclusion, this thesis provides a basis for understanding how in-situ projection can be used for providing cognitive support at workplaces. It identifies the strengths and weaknesses of in-situ projection for cognitive assistance regarding different user groups. Therefore, the findings of this thesis contribute to the field of using Augmented Reality at the workplace. Overall, this thesis shows that using Augmented Reality for cognitively supporting workers during manual assembly tasks and order picking tasks creates a benefit for the workers when working on cognitively demanding tasks.