面向智能机器人柔性装配系统自主控制的主动视觉感知技术研究

项目名称： 面向智能机器人柔性装配系统自主控制的主动视觉感知技术研究

项目编号： No.61503243

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 自动化技术、计算机技术

项目作者： 谷朝臣

作者单位： 上海交通大学

项目金额： 20万元

中文摘要： 针对制造业变革中出现的装配环境非结构化、装配任务频繁变更、装配操作繁重复杂等趋势，本项目将研究定位在机器人自主装配操作的主动视觉感知技术，旨在通过相关研究进一步提高机器人柔性装配系统的智能性、自主性和适应性。.本项目提出研究基于卷积神经网络的目标特征深度学习方法和基于CAD模型特征的迁移训练方法，实现非结构化操作环境下装配目标的主动识别；研究基于改进ASIFT算法的虚实特征点匹配方法和基于正交迭代-卡尔曼滤波的局部特征法向找正方法，实现复杂外形装配目标的位姿重构；研究基于“时序动作流”的感知-操作统一任务建模方法和基于“多元约束模型”的异构约束解析方法，实现主动视觉感知和自主装配操作的任务-信息协同。.项目实施意义包括：从理论层面，将丰富复杂环境下机器人主动视觉感知理论，提升工业机器人的自主控制水平；从应用层面，将推动机器人自主柔性装配技术发展，为工业4.0智能工厂装配自动化提供技术参考。

中文关键词： 自主控制；主动视觉；任务规划；机器人柔性装配

英文摘要： This project focuses on active visual perception technology in robotic autonomous assembly operations, to cope with the trends of unstructured assembly environments, frequent-changing assembly tasks and heavy complex assembly operations in current manufacturing revolution. It is aimed at enhancing the autonomy, intelligence and adaptability of robotic flexible assembly system..This project will investigate convolution neural network based of objects feature deep learning and CAD model feature based transfer training for active recognition of assembly objects in unstructured operation environments; will investigate improved ASFIT algorithm based feature points registration of virtual and real scene and Orthogonal iteration - Kalman filter based local feature normal direction detection for pose reconstruction of assembly object with complex shapes; will investigate “sequential action flow” based unified task modelling of visual perception and assembly operation and “multiple constraint model” based heterogeneous constraint parsing for tasks-information cooperation between active visual perception and autonomous assembly operations..The significance of this project includes: from the theoretical level, this project will contribute to robotic active visual perception theory in complex environment and enhance autonomous control level of industry robots; from application level, this project will promote the autonomous robotic flexible assembly technology toward development and provide technical reference for assembly automation in smart factory of Industry 4.0.

英文关键词： Autonomous Control;Active Vision;Tasks Planning;Robotic Flexible Assembly