摘要增强现实(Augmented Reality)技术用于飞机装配中可以有效提高装配效率。通过无线网络的使用,增强现实设备可以便捷访问各种人工智能服务,将自然场景下的光学字符识别(Optical Character Recognition)技术作为云服务用于 AR 辅助装配中减少人工录入以加速装配过程。针对 AR 辅助飞机线缆装配中线缆识别存在错误的问题,基于贝叶斯纠错器,结合航空线缆规则提出一种数据纠错的后处理流程,使用噪声通道模型对线缆的识别结果数据进行推理纠错,并构建了基于云服务的 AR 辅助线缆装配体系。在该系统的环境下,进行 AR 辅助的航空线缆装配试验,将导线识别准确率从 88% 提高到 98%,验证了该系统具有一定高效性的同时保证了正确性。
Abstract:Augmented reality (AR) technology can effectively improve the efficiency of aircraft assembly. Through the use of wireless networks, augmented reality devices can conveniently access various artificial intelligence services, and use optical character recognition technology in natural scenes as a cloud service for AR assistant assembly to reduce manual entry. Aiming at problem of cable recognition errors in AR assistant aircraft cable assembly, this paper proposes a postprocessing process for data error correction based on Bayesian error correctors and aviation cable rules, and uses a noise channel model to identify cables. The result data was inferred and corrected, and an AR assistant cable assembly system based on cloud services was constructed. Under the environment of this system, the AR assistant aviation cable assembly experiment was conducted to improve the accuracy of cable recognition from 88% to 98%, which verified that the system has certain efficiency and guaranteed correctness.