基于下肢表面肌电的驾驶员紧急制动行为识别
|
摘要
表面肌电信号(surface Electromyography,sEMG)是人体运动检测的主要信息源之一,已被广泛应用.将下肢sEMG信号引入到驾驶员紧急制动行为识别模型中,提取sEMG的时域、频域和模型参数特征组成特征向量.另外,为了提高紧急制动行为的识别准确度,同时采集与其特征相似的常规制动和加速换挡行为数据,利用基于有向无环图的支持向量机构建分类器对三种驾驶行为进行识别.结果表明,同个体紧急制动行为识别率高达90.9%,不同个体识别率平均达81.8%;且该系统能够在紧急操作制动踏板前220ms进行识别,以100km/h的行驶速度计算,制动距离将减少6.1m.
Surface electromyography(sEMG)is one of the main information sources of human motion detection and has been widely used.The lower limb sEMG signal was introduced into the recognition model of drivers' emergency braking behavior,and the features from the time domain,frequency domain and model parameters were extracted to construct the feature vector.In addition,in order to improve the recognition accuracy of emergency braking behavior,the data from conventional braking and accelerated shifting behaviors similar to the characteristics of emergency braking was collected,and these three driving behaviors were identified by using support vector machines based on directed acyclic graph.The results show that the recognition rate of emergency braking behavior of the same individual is up to 90.9%,and the recognition rate of different individuals is 81.8% on average.Moreover,the system can detect emergency braking 220 ms earlier than operating brake pedal.At 100 km/h driving speed,the braking distance will be reduced by 6.1 m.
Surface electromyography(sEMG)is one of the main information sources of human motion detection and has been widely used.The lower limb sEMG signal was introduced into the recognition model of drivers' emergency braking behavior,and the features from the time domain,frequency domain and model parameters were extracted to construct the feature vector.In addition,in order to improve the recognition accuracy of emergency braking behavior,the data from conventional braking and accelerated shifting behaviors similar to the characteristics of emergency braking was collected,and these three driving behaviors were identified by using support vector machines based on directed acyclic graph.The results show that the recognition rate of emergency braking behavior of the same individual is up to 90.9%,and the recognition rate of different individuals is 81.8% on average.Moreover,the system can detect emergency braking 220 ms earlier than operating brake pedal.At 100 km/h driving speed,the braking distance will be reduced by 6.1 m.
引文
[1]文江辉,郭浩,吴超仲,等.基于结构方程模型的疲劳驾驶行为影响因素间量化关系研究[J].交通信息与安全,2014(5):95-102.
[2]弓馨.基于模糊逻辑的驾驶员制动意图辨识方法研究[D].长春:吉林大学,2014.
[3]崔高健,曲代丽,李绍松,等.驾驶员制动意图辨识技术研究现状[J].机械工程与自动化,2016(4):219-221.
[4]MILLER S.A driver intent-based collision avoidance system[J].Alcohol,2009(1):58-64.
[5]LIU Y,SUN Z,WENBIN J I.Development of composite brake pedal-stroke simulator for electro-hydraulic braking system[J].Construction and Building Materials,2014(2):158-163.
[6]王英范,宁国宝,余卓平.乘用车驾驶员制动意图识别参数的选择[J].汽车工程,2011,33(3):213-216.
[7]张旭.基于表面肌电信号的人体动作识别与交互[D].合肥:中国科学技术大学,2010.
[8]井本成,董海清,陈玉娟,等.基于表面肌电的肘关节运动角度预测[J].上海师范大学学报(自然科学版),2017(4):571-576.
[9]胡明霞.表面肌电信号识别特征提取的仿真研究[J].计算机仿真,2013(5):345-348.
[10]薛宁静.多类支持向量机分类器对比研究[J].计算机工程与设计,2011(5):1792-1795.
[2]弓馨.基于模糊逻辑的驾驶员制动意图辨识方法研究[D].长春:吉林大学,2014.
[3]崔高健,曲代丽,李绍松,等.驾驶员制动意图辨识技术研究现状[J].机械工程与自动化,2016(4):219-221.
[4]MILLER S.A driver intent-based collision avoidance system[J].Alcohol,2009(1):58-64.
[5]LIU Y,SUN Z,WENBIN J I.Development of composite brake pedal-stroke simulator for electro-hydraulic braking system[J].Construction and Building Materials,2014(2):158-163.
[6]王英范,宁国宝,余卓平.乘用车驾驶员制动意图识别参数的选择[J].汽车工程,2011,33(3):213-216.
[7]张旭.基于表面肌电信号的人体动作识别与交互[D].合肥:中国科学技术大学,2010.
[8]井本成,董海清,陈玉娟,等.基于表面肌电的肘关节运动角度预测[J].上海师范大学学报(自然科学版),2017(4):571-576.
[9]胡明霞.表面肌电信号识别特征提取的仿真研究[J].计算机仿真,2013(5):345-348.
[10]薛宁静.多类支持向量机分类器对比研究[J].计算机工程与设计,2011(5):1792-1795.