Dynamic obstacle avoidance on quadrotors requires low latency. A class of sensors that are particularly suitable for such scenarios are event cameras. In this work, we present a deep learning-based solution for dodging multiple dynamic obstacles on a quadrotor with a single event camera and onboard 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.
Reference:
N.J. Sanket, C. Parameshwara, C.D. Singh, A. Kuruttukulam, C. Fermüller, D. Scaramuzza, Y. Aloimonos
EVDodgeNet: Deep Dynamic Obstacle Dodging with Event Cameras
IEEE International Conference on Robotics and Automation (ICRA) 2020
PDF: https://arxiv.org/abs/1906.02919
Project Webpage, Software, and Datasets:
http://prg.cs.umd.edu/EVDodgeNet
Affiliations:
N.J. Sanket, C. Parameshwara, C.D. Singh, A. Kuruttukulam, C. Fermüller, Y. Aloimonos are with Perception and Robotics Group (PRG) at the University of Maryland, College Park
http://prg.cs.umd.edu
D. Scaramuzza is with the Robotics and Perception Group, Dep. of Informatics, University of Zurich, and Dep. of Neuroinformatics, University of Zurich and ETH Zurich, Switzerland
http://rpg.ifi.uzh.ch
