加权指数损失下长短时记忆网络换道意图识别模型  

A Long Short-term Memory Intent Recognition Model for Lane Change Based on Time-weighted Exponential Loss

作　　者：王皓昕 李振龙 赵晓华 WANG Hao-xin;LI Zhen-long;ZHAO Xiao-hua(Beijing Key Laboratory of Traffic Engineering,Beijing University of Technology,Beijing 100124,China)

机构地区：[1]北京工业大学城市交通学院交通工程北京市重点实验室,北京100124

出　　处：《科学技术与工程》2021年第1期254-259,共6页Science Technology and Engineering

基　　金：国家自然科学基金(61876011)。

摘　　要：针对车道变换意图识别中数据源单一,传统序列模型难以捕获长序列范围内换道意图且存在长期依赖问题,提出一种结合时间信息加权指数损失函数的长短时记忆(long short-term memory,LSTM)车辆换道意图识别模型。首先,利用驾驶模拟舱、眼动仪进行高速公路驾驶实验,采集车辆运行数据和驾驶员眼动数据;然后,基于LSTM结构单元构建高速公路环境下车辆换道意图识别模型,提出基于时间信息加权的指数损失函数对模型权重进行优化;最后,利用车辆运行数据和驾驶员眼动数据对所提模型加以验证并与其他模型进行对比,所提模型换道识别的准确率为91.33%,宏平均精确率为89.04%,宏平均召回率为92.84%,宏平均F1值为90.33%。结果表明,长短时记忆网络对于长序列换道意图识别过程具有较好的分辨能力,提出的损失函数对模型权重优化具有良好的效果。Considering the single data source in lane change intent recognition,the difficulty to capture the lane change intent in the long sequence range by the traditional sequence model,and the long-term dependency problem,a long short-term memory(LSTM)intent recognition for vehicle lane change model combined with time information weighted exponential loss function was proposed.First,the driving simulator and eye tracker were selected to carry out the highway driving experiment.The driving data and the driver’s eye movement data were collected respectively.Second,based on the LSTM structural unit,the intent recognition model of lane changing under the highway environment was constructed,and the multi-exponential loss function based on time information weighting was proposed to optimize the weights.Finally,the proposed model was verified by vehicle driving data,eye movement data and compared with other models.The accuracy of lane-changing recognition of the proposed model is 91.33%.The macro-precision is 89.04%.The macro-recall is 92.84%.The macro-F1 value is 90.33%.The results show that the long short-term memory network works well for the long-sequence lane change intent recognition process,and the proposed loss function achieves a positive effect on model weight optimization.

关 键 词：智能交通 加权指数损失函数 长短时记忆网络 换道意图识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]