Dynamic obstacle avoidance on quadrotors requires low latency. A class of sensors that are particularly suitable for such scenarios are event cameras. In this paper, we present a deep learning -- based solution for dodging multiple dynamic obstacles on a quadrotor with a single event camera and on-board computation. Our approach uses a series of shallow neural networks for estimating both the ego-motion and the motion of independently moving objects. The networks are trained in simulation and directly transfer to the real world without any fine-tuning or retraining. We successfully evaluate and demonstrate the proposed approach in many real-world experiments with obstacles of different shapes and sizes, achieving an overall success rate of 70% including objects of unknown shape and a low light testing scenario. To our knowledge, this is the first deep learning -- based solution to the problem of dynamic obstacle avoidance using event cameras on a quadrotor. Finally, we also extend our work to the pursuit task by merely reversing the control policy, proving that our navigation stack can cater to different scenarios.
@article{arxiv.1906.02919,
title = {EVDodgeNet: Deep Dynamic Obstacle Dodging with Event Cameras},
author = {Nitin J. Sanket and Chethan M. Parameshwara and Chahat Deep Singh and Ashwin V. Kuruttukulam and Cornelia Fermüller and Davide Scaramuzza and Yiannis Aloimonos},
journal= {arXiv preprint arXiv:1906.02919},
year = {2020}
}
Comments
15 pages, 16 figures, Code and Video can be found at: https://prg.cs.umd.edu/EVDodgeNet