A framework for 3D tracking of frontal dynamic objects in autonomous cars

作者：

Highlights：

• Theoretical development is given for a switched SDRE filter in discrete time domain.

• Generalizing the traditional SFM model by adding a new observation model.

• Combining classic nonlinear estimators with deep learning techniques in practice.

• A real-time performance based on a multi-thread framework.

• Estimation of longitudinal and lateral distances to objects in a dynamic environment.

摘要

•Theoretical development is given for a switched SDRE filter in discrete time domain.•Generalizing the traditional SFM model by adding a new observation model.•Combining classic nonlinear estimators with deep learning techniques in practice.•A real-time performance based on a multi-thread framework.•Estimation of longitudinal and lateral distances to objects in a dynamic environment.

论文关键词：Deep learning,Recognition and 3D tracking,Frontal dynamic objects,Structure from motion,Switched SDRE filter

论文评审过程：Received 27 June 2020, Revised 29 March 2021, Accepted 2 June 2021, Available online 17 June 2021, Version of Record 22 June 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.115343