利用开关卡尔曼滤波器的目标跟踪技术研究

为了使机器人视觉伺服控制系统的目标跟踪精度得到进一步提高,构建一种基于开关卡尔曼滤波器的视觉伺服控制系统。研究视觉伺服的目标跟踪原理,推导相关的数学模型,并分析跟踪误差产生的原因;针对图像采集和处理引入的延时问题,通过卡尔曼滤波估计得到目标运动的速度信息,以此作为前馈量输入视觉伺服控制器,补偿由于目标运动和延时造成的跟踪误差;为了解决卡尔曼滤波器由于目标运动的突然变化而降低估计性能的问题,引入运动监视器以在目标运动突然变化时发出开关信号并重置卡尔曼滤波器;最后对该算法进行实验与仿真。结果表明:基于卡尔曼滤波器的视觉伺服控制器能把跟踪误差控制在1 mm以内,而开关卡尔曼滤波器能有效地减少因目标运动状态突然变化而产生的跟踪误差。

Research on Target Tracking Technology Based on Switched Kalman Filter

In order to improve the target tracking accuracy of the robot visual servo control system,a visual servo control system based on switched Kalman filter was constructed.The principle of visual servo target tracking was studied.The related mathematical model was deduced,and the reason for the tracking error was analyzed.Aiming at the problem of delay introduced by the image acquisition and processing,the velocity of the target was estimated by Kalman filter,and the data was fed forward to the visual servo controller to compensate the tracking error caused by the target motion and delay.For the problem that the performance of the Kalman filter was degraded due to a sudden change of the target motion,a motion monitor was introduced to send a switching signal and reset the Kalman filter when the target motion was suddenly changed.Finally,the algorithm was verified by simulation and experiment.The results indicate that the visual servo system based on Kalman filter can control the tracking error within 1 mm.The switched Kalman filter can effectively decrease the tracking error caused by abrupt motions of target.