Monocular vision-based time-to-collision estimation for small drones by domain adaptation of simulated images

作者：

Highlights：

• A vision-based time-to-collision estimation algorithm using deep learning.

• Convert simulated images into real-like synthetic images using sim-to-real method.

• Proposed an uncertainty and times-series characteristic aware neural network model.

• Verified the proposed model in various indoor environments.

• Compared the performance with state-of-the-art algorithms.

摘要

•A vision-based time-to-collision estimation algorithm using deep learning.•Convert simulated images into real-like synthetic images using sim-to-real method.•Proposed an uncertainty and times-series characteristic aware neural network model.•Verified the proposed model in various indoor environments.•Compared the performance with state-of-the-art algorithms.

论文关键词：Time-to-collision estimation,Aleatoric uncertainty,Epistemic uncertainty,Monte Carlo dropout,Convolutional LSTM,Navigation decision making,Vision-based approach

论文评审过程：Received 10 July 2021, Revised 30 December 2021, Accepted 22 March 2022, Available online 4 April 2022, Version of Record 12 April 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.116973