项目名称: 未校准环境下机器人自适应手眼视觉跟踪研究
项目编号: No.61203361
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 自动化学科
项目作者: 梁新武
作者单位: 上海交通大学
项目金额: 24万元
中文摘要: 视觉跟踪是一种通过摄像机提供的目标图像信息对机器人进行反馈控制使其沿着期望轨迹运动的重要方法。现有手眼视觉跟踪大都基于运动学并假设摄像机内外参数或目标模型已知,无法在摄像机内外参数未校准或目标模型未知的高速运动场合取得理想性能。本项目将对摄像机内外参数和目标几何模型、运动模型均未知下基于动力学的手眼视觉跟踪展开研究。针对固定目标,设计有效的视觉跟踪策略以使机器人沿着由图像特征定义的期望轨迹靠近目标。为了跟踪运动的目标,提出有效的视觉跟踪算法对机器人运动实施控制以使目标图像特征始终保持在图像平面的期望位置上,实现目标锁定或位姿跟踪。考虑到摄像机视场约束问题的重要性,将研究摄像机视场约束下的视觉跟踪方法。为了解决图像速度测量噪声较大的问题,将设计不使用速度的视觉跟踪算法。同时基于机器人的非线性动力学分析系统稳定性和跟踪误差收敛性。最后通过六自由度机器人手眼视觉跟踪实验验证所提跟踪算法的有效性。
中文关键词: 视觉伺服;视觉跟踪;自适应控制;机器人;未校准环境
英文摘要: Visual tracking is an important approach that uses image information of objects from a camera for feedback control of robots such that the robot can follow its desired trajectories. Most of the existing eye-in-hand visual tracking approaches are based on the robot kinematics and under the assumption that the camera intrinsic and extrinsic parameters or the object model are already known, and hence cannot obtain satisfying performance for high-speed motion control when the camera parmameters are uncalibrated or the object model is unknown. In this project, we will investigate the dynamics-based eye-in-hand visual tracking problem under the assumption that the camera intrinsic and extrinsic parameters and the geometric and motion model of objects are all unknown. For static objects, we will design effective visual tracking strategies to ensure that the robot can approach the object along the desired trajectories defined by the image features. To track moving objects, we will propose effective visual tracking algorithms to control the motion of the robot such that image features of the moving object will always remain at their respective desired positions on the image plane for locking the moving object or pose tracking. Considering that the problem of the camera's field of view constraint is very important, we wil
英文关键词: visual servoing;visual tracking;adaptive control;robot;uncalibrated environments