Edge Computing Driven Scene-aware Intelligent Augmented Reality for Manual Assembly

Abstract Wearable augmented reality (AR) can superimpose virtual models or annotation on real scenes, and which can be utilized in assembly tasks and resulted in high-efficiency and error-avoided manual operations. Nevertheless, most of existing AR-aided assembly operations are based on the predefined visual instruction step-by-step, lacking scene-aware generation for the assembly assistance. To facilitate a friendly AR-aided assembly process, this paper proposed an Edge Computing driven Scene-aware Intelligent AR Assembly (EC-SIARA) system, and smart and worker-centered assistance is available to provide intuitive visual guidance with less cognitive load. In beginning, the connection between the wearable AR glasses and edge computing system is established, which can alleviate the computation burden for the resource-constraint wearable AR glasses, resulting in a high-efficiency deep learning module for scene awareness during the manual assembly process. And then, based on context understanding of the current assembly status, the corresponding augmented instructions can be triggered accordingly, avoiding the operator’s cognitive load to strictly follow the predefined procedure. Finally, quantitative and qualitative experiments are carried out to evaluate the EC-SIARA system, and experimental results show that the proposed method can realize a worker-center AR assembly process, which can improve the assembly efficiency and reduce the occurrence of assembly errors effectively.

Download Full-text

BricksAR: exploring the potential of augmented reality to inform assembly processes in construction

10.5753/svr_estendido.2020.12953 ◽

2020 ◽

Author(s):

João Pedro Andrade Caixeta ◽

André Luís De Araújo

Keyword(s):

Augmented Reality ◽

Real World ◽

Low Cost ◽

Manual Assembly ◽

Ongoing Research ◽

The Real ◽

Modeling Simulation ◽

Design Production ◽

Manufacturing Methods ◽

Assembly Operations

The use of Augmented Reality (AR) systems in construction processes can represent an essential transformation in the communication between design and production. However, supposing that design-production translations can be obtained from several manufacturing methods (such as robotic, manual, modular, non-modular, and others), there are not enough studies that explored the potential uses of AR as an assistant for assembly operations. To tackle this problem, this ongoing research proposes to investigate the potentials of a low-cost and marker-based AR system to conduct different manual assembly processes. With the observational focus on aspects of precision and feasibility, we used scientific reductions based on modeling, simulation, and prototyping to provide inferences about the proposed tool's behavior in the real world.

Download Full-text

Real Time Implementation of Learning-Forgetting Models for Cycle Time Predictions of Manual Assembly Tasks after a Break

Sustainability ◽

10.3390/su12145543 ◽

2020 ◽

Vol 12 (14) ◽

pp. 5543

Author(s):

Steven Hoedt ◽

Arno Claeys ◽

El-Houssaine Aghezzaf ◽

Johannes Cottyn

Keyword(s):

Real Time ◽

Cycle Time ◽

Mixed Model ◽

Manufacturing Companies ◽

Manual Assembly ◽

Time Prediction ◽

Curve Model ◽

Assembly Operations ◽

Assembly Tasks ◽

Mixed Model Assembly

Industry 4.0 provides a tremendous potential of data from the work floor. For manufacturing companies, these data can be very useful in order to support assembly operators. In literature, a lot of contributions can be found that present models to describe both the learning and forgetting effect of manual assembly operations. In this study, different existing models were compared in order to predict the cycle time after a break. As these models are not created for a real time prediction purpose, some adaptations are presented in order to improve the robustness and efficiency of the models. Results show that the MLFCM (modified learn-forget curve model) and the PID (power integration diffusion) model have the greatest potential. Further research will be performed to test both models and implement contextual factors. In addition, since these models only consider one fixed repetitive task, they don’t target mixed-model assembly operations. The learning and forgetting effect that executing each assembly task has on the other task executions differs based on the job similarity between tasks. Further research opportunities to implement this job similarity are listed.

Download Full-text

A High-Efficiency Spike Sorting Cloud-Edge Computing System with DL-DFCM

2019 International Conference on Internet of Things (iThings) and IEEE Green Computing and Communications (GreenCom) and IEEE Cyber, Physical and Social Computing (CPSCom) and IEEE Smart Data (SmartData) ◽

10.1109/ithings/greencom/cpscom/smartdata.2019.00025 ◽

2019 ◽

Author(s):

Shengyi Qian ◽

Hui Hong ◽

Zhengfei Zhu ◽

Keming Chen ◽

Nenggan Zheng ◽

...

Keyword(s):

High Efficiency ◽

Computing System ◽

Edge Computing ◽

Spike Sorting

Download Full-text

Comparing Augmented Reality-Assisted Assembly Functions—A Case Study on Dougong Structure

Applied Sciences ◽

10.3390/app10103383 ◽

2020 ◽

Vol 10 (10) ◽

pp. 3383 ◽

Cited By ~ 2

Author(s):

Chih-Hsing Chu ◽

Chien-Jung Liao ◽

Shu-Chiang Lin

Keyword(s):

Augmented Reality ◽

Assembly Process ◽

Functional Design ◽

Complex Objects ◽

Manual Assembly ◽

Scale Models ◽

Architectural Innovation ◽

Assembly Instructions

The Dougong structure is an ancient architectural innovation of the East. Its construction method is complex and challenging to understand from drawings. Scale models were developed to preserve this culturally-unique architectural technique by learning through their assembly process. In this work, augmented reality (AR)-based systems that support the manual assembly of the Dougong models with instant interactions were developed. The first objective was to design new AR-assisted functions that overcome existing limitations of paper-based assembly instructions. The second one was to clarify whether or not and how AR can improve the operational efficiency or quality of the manual assembly process through experiments. The experimental data were analyzed with both qualitative and quantitative measures to evaluate the assembly efficiency, accuracy, and workload of these functions. The results revealed essential requirements for improving the functional design of the systems. They also showed the potential of AR as an effective human interfacing technology for assisting the manual assembly of complex objects.

Download Full-text

Evaluation of an Augmented Reality Instruction for a Complex Assembly Task

i-com ◽

10.1515/icom-2021-0005 ◽

2021 ◽

Vol 20 (1) ◽

pp. 63-72

Author(s):

Johannes Funk ◽

Ludger Schmidt

Keyword(s):

Augmented Reality ◽

Knowledge Transfer ◽

Manual Assembly ◽

Complex Assembly ◽

Assembly Task ◽

Rail Vehicles ◽

Instruction Type ◽

Low Costs ◽

Nasa Tlx ◽

Assembly Tasks

Abstract This study compares the use of a marker-based AR instruction with a paper instruction commonly used in manual assembly. Hypotheses were tested as to whether the instruction type affects assembly time, number of errors, usability, and employee strain. Instead of student participants and artificial assembly tasks (e. g. Lego assemblies), the study was conducted with 16 trainees in a real workplace for the assembly of emergency door release handles in rail vehicles. Five assembly runs were performed. Assembly times and assembly errors were determined from recorded videos. Usability (SUS) and strain (NASA-TLX) were recorded with questionnaires. After a slower assembly at the beginning, the AR group assembled significantly faster in the fifth run. The comparable number of errors, usability and strain make marker-based AR applications interesting for knowledge transfer in manual assembly, especially due to the easy entrance and low costs.

Download Full-text