Deep learning-based dynamic object classification using LiDAR point cloud augmented by layer-based accumulation for intelligent vehicles

作者：

Highlights：

• A layer-based registration applies the vehicle’s motion to the registration method.

• Shape information is strengthened by accumulating point clouds of dynamic objects.

• The strengthened shape information improves the classification performance.

• 3D computer aided design models can be used to train a deep learning model.

摘要

•A layer-based registration applies the vehicle’s motion to the registration method.•Shape information is strengthened by accumulating point clouds of dynamic objects.•The strengthened shape information improves the classification performance.•3D computer aided design models can be used to train a deep learning model.

论文关键词：LiDAR,Registration,Deep learning,Dynamic object classification,Intelligent vehicle

论文评审过程：Received 9 June 2019, Revised 3 July 2020, Accepted 6 August 2020, Available online 20 August 2020, Version of Record 10 February 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.113861