基于图像序列拓扑关系的扎针机器人控制系统设计

Design of control system of needle robot based on topological relation of image sequence

摘要:

由于当前扎针机器人定位参数存在误差，导致扎针成功率低，造成患者要忍受多次痛苦的问题，设计了一种扎针机器人控制系统。该系统由机械臂部分、上位机控制系统、高清激光图像采集系统、和信号传感器等部分组成，主要硬件部分选用DHT01型医用高清镜头、PCI8814图像采集卡、和性能更强的ATMEGA32L型单片机，保障采集图片的清晰度和对机械臂的控制精度。在图像序列处理算法上计算多帧图像的相似度，使用FAST算子提取图像的特征点，并基于关键帧优化图像序列的拓扑关系，实现目标位置的准确定位。实验结果表明，提出扎针机器人系统设计在3个轴向的偏差度平均值分别为0.0449、0.0383、和0.0387，趋近于理论上的中心位置，扎针一次成功率可以达到94.50%。

Abstract:

Due to the error in the positioning parameters of current needle stick robots, the success rate of needle sticks is low, which causes the patient to endure many painful problems. A needle stick robot control system is designed. The system consists of a robot arm, a host computer control system, a high-definition laser image acquisition system, and a signal sensor. The main hardware part is selected DHT01 medical high-definition lens, PCI8814 image acquisition card, and a stronger ATMEGA32L single-chip microcomputer to ensure The sharpness of the collected pictures and the control precision of the robotic arm. The similarity of multi-frame images is calculated on the image sequence processing algorithm, the feature points of the images are extracted using the FAST operator, and the topological relationship of the image sequence is optimized based on the key frames to achieve accurate positioning of the target position. The experimental results show that the average deviation of the three-axis robot system design is 0.0449, 0.0383, and 0.0387, which approach the theoretical center position, and the success rate of one-time needle insertion can reach 94.50%.