救援机器人的非接触式人体体征视觉测量方法研究

Vital signs non-contact measurement method by rescue robot

目的:研究救援机器人的非接触式人体体征测量方法,实现救援机器人低耗、快速地获取伤员的心率、呼吸率体征,提升救援机器人自主搜寻的智能化水平。方法:以远程光电容积脉搏波描记法为基础,通过KLT(Kanede-Lucas-Tomasi)跟踪算法对人脸部位进行实时跟踪;采用最大比合并算法分集合并感兴趣区不同像素区域的脉搏波信号,获取更具鲁棒性的脉搏波信号用于心率、呼吸率的测量。以地面移动机器人作为载体平台开展模拟伤员体征测量实验测试,将实验得到的心率、呼吸率数据作为预测值;采用EQ02+Life Monitor医用生理监测背心作为对照设备,将其采集的心理信号作为对照值。结果:与对照设备的测量值比较,该方法的总体心率测量值平均误差为1.19次/min,均方根误差为1.58次/min,相关性系数r=0.97;总体呼吸率测量值平均误差为1.63次/min,均方根误差为1.92次/min,相关性系数r=0.85。结论:该方法能有效实现远程获取人体体征,与医用生理监测背心同步采集的数据相比具有很好的一致性,且准确度较高,为进一步提升救援机器人执行伤员救援任务时的实战性能打下了基础。

Objective To explore the non-contact measurement method by the rescue robot for vital signs such as heart rate and respiratory rate to enhance auto search intellectualization of the rescue robot.Methods Based on remote photoplethy smography(RPPG),real-time face tracing was realized with Kanede-Lucas-Tomasi tracking algorithm;maximum ratio combining algorithm was used to execute diversity combining of the pulse wave signals at different ROI pixel areas,and then the obtained pulse wave signals with high robustness were applied to measuring the heart rate and respiratory rate.The ground robot was involved in simulated patient vital signs measurement as the carrier platform to obtain the predictive values for heart rate and respiratory rate,and EQ02+Life Monitor medical physiological monitoring vest was used to acquire the control values for physiological signals.Results When compared with the control values,the measured values for heart rate had the mean error being 1.19 times per minute,root mean square error being 1.58 times per minute and correlation coefficient r being 0.97;the measured values for respiratory rate had the mean error being 1.63 times per minute,root mean square error being 1.92 times per minute and correlation coefficient r being 0.85.Conclusion The proposed method can effectively obtain human vital signs remotely,also has high consistency andaccuracy when compared with medical physiological monitoring vest,so that the practicability of the rescue robot may be enhanced further.[Chinese Medical Equipment Journal,2020,41(6):26-31]