模糊自适应卡尔曼滤波技术在球果采集机器人控制系统中的应用

Application of Fuzzy Self- Adapting Kalman Filter in Control System of Pinecone Picking Robot

为实现自动采果,采用基于模糊自适应卡尔曼滤波的RBF神经网络对机械手动作进行控制.通过MATLAB编程固化到芯片中,将在线获得的三维激光扫描仪及传感器数据进行实时处理并控制采果运动.试验表明：采用的神经网络控制系统工作有效,采果机器人每天可采落叶松球果700～1000 kg,效率为未采用神经网络控制时的1.4～2.0倍,为人工采集的40～60倍.

In order to pick cones automatically,RBF neural network based on fuzzy self-adapting Kalman filter is applied to control the manipulator motion of robot. By programming with MATLAB and solidifying the program to a chip,the data obtained from the three-dimension laser scanner and the sensors on line are processed so as to control the operation of picking cone automatically. The test shows that the automatic control system of RBF neural network is effective and the cones picked by the robot is about 700～1 000 kg per day,its efficiency is about 1 . 4～2 . 0 times than that of the robot without RBF control system and about 40~60 times than that of a picker by hand.