Abstract: To simplify the networking processes of ICVs before multi-sensor fusion, a networking model of self-search—self-identification—self-calibration(S-SIC) was proposed for a PnP environmental sensing sensor. The PnP sensors were searched and connected to ADAS domain by using the depth-first search planning algorithm. The PnP sensors were identified and initialized by using the broadcast messages. The PnP sensors were calibrated by the coordinate transformation matrix of the vehicle body coordinate system and compensation algorithm. Aiming at the incompatibility of ICV multi-sensor data-level and feature-level fusion, a K-m.AW decision-level fusion algorithm was proposed. The experimental results show that the S-SIC model search success rate reaches 92%, and the average search time is as 1.79 s. In the trajectory estimation of the preceding vehicle, the multi-scene fusion estimation based on K-m.AW algorithm is increased by 7.6% and 11.8% respectively compared with contrast algorithms.

Key words:  , intelligent connected vehicle(ICV), multi-sensor, plug-and-play(PnP), information fusion, clustering

摘要： 针对智能网联车（ICV）多传感器融合前组网流程繁琐的问题，提出了一种车载即插即用（PnP）环境感知传感器“自搜索(SS)自识别(SI)自标定（SC）”组网模型（简称“S-SIC”）。采用深度优先搜索规划算法自搜索PnP传感器并接入ADAS域；采用广播报文自识别PnP传感器并进行初始化；采用车体坐标系坐标转换矩阵及补偿算法自标定PnP传感器。针对ICV多传感器数据及特征级融合时设备难以兼容的问题，提出了K-m.AW决策级融合算法。实验结果表明：S-SIC模型搜索成功率达92%，搜索平均时间为1.79 s；在前车轨迹估计中，基于K-m.AW算法的多场景融合估计准确率较对比算法分别提高了7.6%、11.8%。

关键词: 智能网联车, 多传感器, 即插即用, 信息融合, 聚类