摘要
针对人工智能相关专业深度机器学习理论与实践脱节的情况,结合市场对垃圾分拣的实际需求,设计了一款基于TensorFlow的垃圾视觉分拣机器人实验平台。实验平台硬件部分包括:NVIDIA Jetson TX2嵌入式机器学习平台、智能机械臂控制系统;软件部分基于TensorFlow深度学习框架,通过数据采集、数据标注、MobileNet SSD模型的迁移训练等若干步骤,实现了垃圾的视觉分拣。实验平台可满足深度学习、工业机器人、机器视觉等先进技术的实践教学要求,有助于提高学生的创新与实践能力。
To solve the disconnection between the theory and practice of deep machine learning in AI related specialties and the actual demand of garbage sorting in the market,an experimental platform of garbage visual sorting robot based on TensorFlow was designed.The hardware part of the experimental platform includes NVIDIA Jetson TX2 embedded machine learning platform,intelligent manipulator control system,and the software part is based on TensorFlow deep learning framework.Through data acquisition,data annotation,mobile Net SSD model migration training and other steps,the visual sorting of garbage is achieved.The experimental platform can meet the teaching requirements for advanced technologies such as deep learning,industrial robots,machine vision,and help to improve students’innovation and practical ability.
作者
吴蓬勃
姚美菱
王拓
孙青华
WU Pengbo;YAO Meiling;WANG Tuo;SUN Qinghua(Department of Telecommunication Engineering,Shijiazhuang Post&Communication Technology Institute,Shijiazhuang 050021,China)
出处
《实验室研究与探索》
CAS
北大核心
2020年第6期117-122,共6页
Research and Exploration In Laboratory
基金
河北省高等学校科学技术研究青年基金项目(QN2019336)
河北省人力资源与社会保障厅项目(JRS-2019-1143)。