融合多车风格感知与交互特征的换道行为预测

韩泽一,王文璇,王元庆

Prediction of lane-changing behavior integrating multi-vehicle style perception and interaction features

Zeyi HAN,Wenxuan WANG,Yuanqing WANG

表 9 2.0 s窗口时间下前20个贡献度最高的特征值

Tab.9 Top 20 features with highest contribution values under 2.0-second window time

Rank No. $\overline{\rm{SHAP}} $值 Co/% Rank No. $\overline{\rm{SHAP}} $值 Co/% 1 1 1.279 19.38 11 23 0.159 2.41 2 43 0.941 14.26 12 6 0.159 2.41 3 25 0.624 9.46 13 80 0.143 2.17 4 32 0.383 5.82 14 44 0.141 2.14 5 27 0.310 4.70 15 126 0.122 1.86 6 127 0.287 4.36 16 124 0.119 1.81 7 2 0.236 3.59 17 67 0.092 1.39 8 69 0.221 3.36 18 125 0.073 1.11 9 86 0.179 2.72 19 122 0.072 1.10 10 87 0.169 2.56 20 70 0.066 1.00