摘要
人机共驾中,共驾模式的选择和驾驶控制权的分配高度依赖于对驾驶人状态的正确识别。为了分析人机共驾中驾驶人的状态,对行车风险场模型进行重构,通过构建风险场力作用机制,建立包含驾驶人特性、自车特性和外部风险特性的人-车-路闭环系统中的驾驶人风险响应度模型,用于表征驾驶人对风险的认知能力和应对倾向。根据24位驾驶人在跟车和并道2个场景中的驾驶试验结果,对不同风险响应度下驾驶人的驾驶特性进行分析。研究结果表明:驾驶人风险响应度在驾驶过程中具有时变性,在驾驶人个体之间和不同驾驶场景间均存在差异性。在风险响应度分别为低、中、高的3类驾驶片段中,驾驶人在驾驶时的碰撞时间倒数TTCi和加减速行为均具有显著差异(p<0.05);风险响应度较高的保守型驾驶中,驾驶人行车时倾向于保持较小的TTCi(均值为-0.48s-1,标准差为1.25s-1),单位时间内制动操作最多[均值为0.65次·(15s)-1],总体驾驶风格倾向于规避风险;风险响应度较低的激进型驾驶中,驾驶人行车时倾向于保持最大的TTCi(均值为0.28s-1,标准差为0.42s-1),相较于保守型驾驶,单位时间内加速操作较多[均值为0.48次·(15s)-1],制动操作较少[均值为0.50次·(15s)-1],总体驾驶风格倾向于追求行驶效率;风险响应度居中的平衡型驾驶中,驾驶人行车时所保持的TTCi居中(均值为0.04s-1,标准差为0.36s-1),单位时间内加速操作[均值为0.23次·(15s)-1]和制动[均值为0.41次·(15s)-1]操作总数最少,对于行驶效率和行车安全的追求相对均衡。相较于以往将驾驶人作为孤立个体的驾驶人状态评估方法,所提出的驾驶人风险响应度模型可以依据驾驶人在人-车-路交互中的驾驶表现,更为全面地反映驾驶人的个性化驾驶状态。
Proper mode selection and driving-authority allocation in man-machine shared driving depends highly on the reliable identification of the driver state.A driving-risk field model was reconstructed to analyze the driver state in man-machine shared driving,and an inter-field mechanism of driving-risk field force was proposed.A driving-risk-response model for a humanvehicle-road close-loop system was thus established to reflect driver cognitive ability and coping tendency towards driving risk,considering the characteristics of external risk,as well as the egocar and its driver.Road experiments involving 24 drivers were conducted in both car-following and cut-in scenarios,and the driving characteristics of drivers with different levels of driving risk response were analyzed.The results show that the driving risk response varies with time during driving,and is distinct for individuals and different driving scenarios.In the three types of driving segments where drivers exhibit low,medium,and high driving risk response,drivers perform significantly differently(p<0.05)in terms of inverse time-to-collision(TTCi)and acceleration/deceleration maneuvers.In conservative driving segments with high driving risk response,the drivers drive with the lowest TTCi(mean is-0.48 s-1,SD is 1.25 s-1)and the most braking maneuvers[mean is 0.65 times· (15 s)-1]to avoid risks.In aggressive driving segments with low driving risk response,drivers maintain the highest TTCi(mean is 0.28 s-1,SD is 0.42 s-1)and accelerate the most [mean is 0.48 times· (15 s)-1]with less braking [mean is 0.50 times· (15 s)-1],pursuing driving efficiency.In balanced driving segments with medium driving risk response,the least acceleration [mean is 0.23 times· (15 s)-1]and deceleration[mean is 0.41 times· (15 s)-1]maneuvers are made,and mediumTTCi(mean is 0.04 s-1,SD is0.36 s-1)is frequently maintained to keep the balance between driving safety and efficiency.Compared with previous driver-state evaluation methods,the isolated consideration of the driver is replaced by the performance of the driver in human-vehicle-road interaction,and thus drivers’ personalized driving states are accurately reflected with the driving-risk-response model.
作者
何仁
赵晓聪
王建强
HE Ren;ZHAO Xiao-cong;WANG Jian-qiang(School of Automotive and Traffic Engineering,Jiangsu University,Zhenjiang 212013,Jiangsu,China;State Key Laboratory of Automotive Safety and Energy,Tsinghua University,Beijing 100084,China)
出处
《中国公路学报》
EI
CAS
CSCD
北大核心
2020年第9期236-250,共15页
China Journal of Highway and Transport
基金
国家杰出青年科学基金项目(51625503)
江苏省研究生科研与实践创新计划项目(SJKY19_2537)。
关键词
交通工程
驾驶人风险响应度
人机共驾
驾驶人状态
行车风险场
traffic engineering
driving risk response
man-machine shared driving
driver state
driving-risk field
