论文部分内容阅读
针对传统的空间机器人遥操作控制方法受时延影响严重的问题,基于最新研制的4自由度空间机器人,提出一种基于多目相机的自主复合控制策略.该策略包含了眼在手和眼到手2种相机配置结构,每个相机均具有独立的位姿计算能力,前者保证了伺服控制的精度以及操作的灵活性,后者能够在观察到全局视景的情况下做出伺服控制.为满足机器人视觉实时控制的需求,对视觉处理算法进行了特别考虑.首先,设计了基于多边形形状拟合的特征识别方法,提出曲线矢量数据贪婪算法来处理图像遍历拟合过程中计算密集的问题;其次,结合特征识别结果和模型目标的空间信息,提出基于弱透视模型的单目位姿估算及优化算法;最后,依据所提策略在实验室环境中完成了自主导航及捕获任务,验证了在较低层次进行这种复合控制的可行性和有效性.
Aiming at the problem that the traditional remote control method of space robot is seriously affected by time delay, this paper proposes a multi-camera based autonomous compound control strategy based on the newly developed 4-DOF space robot.This strategy includes both eye hand and eye hand 2 kinds of camera configuration structure, each camera has its own position and posture computing capabilities, the former guarantees the accuracy of servo control and operational flexibility, the latter can make a servo control in the case of global vision to meet The requirements of robot vision real-time control are specially considered for vision processing algorithms.Firstly, feature recognition method based on polygon shape fitting is designed and curve vector data greedy algorithm is proposed to deal with computationally intensive problem in image traversal fitting process. Secondly, , Combined with the results of feature recognition and the spatial information of the model object, a monocular pose estimation and optimization algorithm based on the weak perspective model is proposed. Finally, based on the proposed strategy, autonomous navigation and acquisition tasks are completed in a laboratory environment, The feasibility and effectiveness of such composite control at a lower level.