基于卷积神经网络的手术机器人控制系统设计

针对传统系统控制精准度低的问题，提出了基于卷积神经网络的手术机器人控制系统设计。根据基于卷积神经网络的手术机器人控制原理，设计控制系统总体结构，选用PCI插槽上直接内插CAN适配卡作为上位机核心组件，采用C++编写的Lib库和DLL库为驱动程序提供适配卡。通过下位机三个节点，处理相关信号，并进行量程转换和越限判断，确保机器人不会失控。选用80C592微控制器设计关节驱动节点结构，以高速工作方式向控制器提供向总线的差动发送和接受能力，避免外界干扰。设计基于视觉的持镜臂，为手术过程提供上下、左右、前后的运动的手术视野。分别采用FN3002力传感器和MPS-M拉线式位移传感器获取相关传感数据，采用卷积神经网络深度学习方法，设计持镜臂运动控制步骤，采用VC++6.0工具，控制软件程序，避免抖动或者误操作主手现象的出现。由实验结果可知，该系统持镜臂轨迹规划与期望轨迹一致，简化了控制系统的复杂性。

Design of Surgical Robot Control System Based on Convolutional Neural Network

Aiming at the problem of low control accuracy of traditional systems, a surgical robot control system design based on convolutional neural networks is proposed. According to the control principle of surgical robot based on convolutional neural network, the overall structure of the control system is designed. The CAN adapter card is directly inserted into the PCI slot as the core component of the host computer. The Lib library and DLL library written in C ++ are used to provide suitable driver programs. With card. Through the three nodes of the lower computer, the relevant signals are processed, and the range conversion and limit violation judgment are performed to ensure that the robot will not lose control. The 80C592 microcontroller is selected to design the joint drive node structure, and the controller is provided with a differential transmission and reception capability to the bus in a high-speed operation mode to avoid external interference. The vision-based lens-holding arm is designed to provide a surgical field of vision for up, down, left and right, and back and forth movements during the operation. FN3002 force sensor and MPS-M pull-type displacement sensor were used to obtain relevant sensing data. Convolutional neural network deep learning method was used to design the lens arm motion control steps. VC ++ 6.0 tools were used to control software programs to avoid jitter or The appearance of mishandling the main hand. From the experimental results, it can be seen that the trajectory planning of the mirror holding arm of the system is consistent with the expected trajectory, which simplifies the complexity of the control system.