A-STAR: Aerospace Standardisation Technology for Assembly and Repair

在我们的A-STAR中\*（航空航天装配和维修标准化技术）项目我们的目标是通过专注于VIOLET的开发和演示来提高航空航天制造业的生产力、材料和能源效率：VIOLET是一种基于先进计算机视觉的创新人工智能（AI）装配助手，能够自动检查数字制造路线卡：* 自动缺陷检测，从而减少过程中的装配错误 * 通过自动生成的历史记录实现可追溯性和合规性 * 在工厂车间进行自动技能采集，用于操作员培训和知识传播该项目将从关注材料和能源开始，以英国最大的航空轮胎制造商Dunlop Aircraft Tyres为主要示范企业，进行密集的航空组装流程，然后通过在Brookhouse Aerospace的部署和测试，展示航空可扩展性，一家领先的民用和国防复合材料结构制造商。A-STAR将利用最先进的先进的AI计算机视觉识别操作员在装配过程中的操作，以确保满足严格的航空航天质量标准。与传统的自动光学检测相比，卷积神经网络的使用为模式识别提供了更多的通用性，并且在检测异常方面表现更好。然后，对产品执行的操作的数字记录可以存储为质量管理的证明，并用于培训新的特殊工艺操作员。该记录可用于数字连接工厂，以便通过供应链跟踪缺陷并用于证明质量标准。我们正在超越最先进的计算机视觉和人类行为识别技术进行创新，将人工智能应用于航空航天手动流程中的人类行为数据，并从长远来看推动人们在制造业中的工作方式。交付后，A-STAR将成为人工智能辅助装配的模板，实现能源和材料的减少，并实现可持续的能源弹性装配和维修操作。

Within our A-STAR\* (Aerospace Standardisation Technology for Assembly and Repair) project we aim to deliver productivity, material and energy efficiency improvements in aerospace manufacturing by focusing on the development and demonstration of VIOLET: an innovative Artificial Intelligence (AI) assembly assistant based on advanced computer vision, capable of automatically checking digital manufacture route-cards:* automatic defect detection thereby reducing the in-process assembly errors* traceability & compliance via automatically generated historical records* automatic skill capture on the factory floor for operator training and knowledge disseminationThe project will start with a focus on material and energy-intensive aerospace assembly processes with Dunlop Aircraft Tyres as the lead demonstrator - the UK's largest Aerospace Tyre Manufacturer - and then demonstrate aerospace scalability through deployment and testing at Brookhouse Aerospace, a leading manufacturer of composite structures for civil and defence.A-STAR will leverage state-of-the-art AI computer vision to recognise operator actions on assembly processes to ensure that stringent aerospace quality standards are met. The use of convolutional neural networks offers more generalisability for pattern recognition and performs better for detecting anomalies compared to traditional automated optical inspection. A digital record of the actions that have been performed on a product can then be stored as a proof of quality management as well as be used to train new special-process operators. This record is useful for digitally connecting factories so that defects can be traced through supply-chains and used to prove quality standards. We are innovating beyond state-of-the-art computer vision and human action recognition, bringing AI to human action data in aerospace manual processes and in the long-term advancing the way in which people work in manufacturing.When delivered, A-STAR will become the template for AI-assisted assembly, delivering energy and material reduction and enable a sustainable, energy-resilient approach to assembly and repair operations.