摘要
机器人视觉伺服系统是机器人研究领域的一个重要研究方向。以图像为基础的视觉伺服机器人模型中,有许多的不确定性,如机器人动力学模型,运动学模型,摄像机系统以及雅可比矩阵等。多数的视觉伺服系统往往只考虑某一、两个部分,而其它的不确定性依然影响定位目标的精度。在使用了神经网络的控制器中加入卡尔曼滤波,在摄像机与机器人坐标系无标定的情况下,对雅可比矩阵进行实时在线估计,从而提高了视觉伺服系统的精确性。仿真验证了本算法的可行性和有效性。
Robotic visual servo system is an important subject in the field of robots.Image-based visual servo model has a lot of uncertainties,for example,robot dynamics model,robot kinematics model,camera system and Jacobian matrix and so on.Most of visual servo methods always considered one or two aspects,but other uncertainties still affected the accuracy of tracking the target for the robot.Based on the traditional controller with neural network,a new control algorithm was provided in the uncalibrated camera and coordination systems with the Kalman filter that estimated Jacobian matrix on-line,thereby improving the accuracy of the system.Simulation result shows the feasibility and effectiveness of the algorithm.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2010年第12期2934-2937,共4页
Journal of System Simulation
关键词
视觉伺服
控制算法
神经网络
雅可比矩阵
卡尔曼滤波
visual servo
control algorithm
neural network
Jacobian matrix
Kalman-bucy filter
