基于特征增稳的混合固态激光雷达目标检测  

Semi Solid-State LiDAR Object Detection Algorithm Enhanced by Feature Stability Enhancement

作　　者：金立生 张洪瑜 郭柏苍 Jin Lisheng;Zhang Hongyu;Guo Baicang(School of Vehicle and Energy,Yanshan University,Qinhuangdao 066004;State Key Laboratory of Crane Technology for Yanshan University,Qinhuangdao 066004;Yanshan University,Hebei Key Laboratory of Special Carrier Equipment,Qinhuangdao 066004)

机构地区：[1]燕山大学车辆与能源学院,秦皇岛066004 [2]燕山大学,起重机械关键技术全国重点实验室,秦皇岛066004 [3]燕山大学,河北省特种运载装备重点实验室,秦皇岛066004

出　　处：《汽车工程》2024年第6期1015-1024,共10页Automotive Engineering

基　　金：国家重点研发计划(2021YFB3202200);国家自然科学基金(52072333,52202503)资助。

摘　　要：稳定的点云特征提取对激光雷达三维目标检测至关重要。针对现有深度学习算法仅能处理机械旋转式激光雷达点云数据,而对混合固态激光雷达数据支持不足的问题,本文基于IA-SSD搭建了适用于混合固态激光雷达的目标检测模型。首先,在点云编码前端添加了Cloth Simulation Filtering(CSF)用于地面次优点过滤;其次,利用局部特征融合和全局双线性正则化层组成的注意力机制,从局部和全局共同推动点云几何信息与特征信息融合;再次,利用GhostGConv替换原有低效的逐点卷积,通过通道混洗机制加强点云的特征交互,构建了增强型特征提取网络。最后,在点检测器IA-SSD中整合了上述模块完成模型构建。在混合固态激光雷达数据集SimoSet开展的验证结果表明,所提出的方法显著优于SimoSet数据集支持的其他算法精度指标;在KITTI数据集中等难度检测中,所提方法将IA-SSD三分类平均检测精度分别提升了1.17、1.47、0.5百分点。Stable point cloud feature extraction is crucial for 3D object detection using LiDAR.For the limitation of existing deep learning algorithms that can only handle point clouds from mechanically rotating LiDAR but lack support for semi solid-state LiDAR data,in this paper a target detection model suitable for semi solid-state LiDAR based on IA-SSD is established.Firstly,Cloth Simulation Filtering(CSF)is added at the front end of point cloud encoding for ground suboptimal filtering.Secondly,an attention mechanism composed of local feature fusion and global bilinear regularization layers is utilized to promote the fusion of geometric information and feature information of point clouds from both local and global perspectives.Thirdly,GhostGConv is employed to replace the original inefficient point-by-point convolution,and enhanced feature interaction of point clouds is achieved through channel shuffling mechanism to construct an enhanced feature extraction network.Finally,the above modules are integrated into the point detector IA-SSD to complete model construction.The validation results conducted on the SimoSet semi solid-state lidar dataset show that the proposed method significantly outperforms other algorithms supported by the SimoSet dataset in terms of accuracy metrics.In the medium difficulty detection tasks on the KITTI dataset,the proposed method enhances the average detection accuracy of IA-SSD in the three categories by 1.17,1.47,and 0.5 percentage points,respectively.

关 键 词：车辆工程 环境感知 目标检测 混合固态激光雷达 特征增稳 

分 类 号：TN958.98[电子电信—信号与信息处理] U495[电子电信—信息与通信工程]