Header

UZH-Logo

Maintenance Infos

EVDodgeNet: Deep Dynamic Obstacle Dodging with Event Cameras


Sanket, Nitin J; Parameshwara, Chethan M; Singh, Chahat Deep; Kuruttukulam, Ashwin V; Fermuller, Cornelia; Scaramuzza, Davide; Aloimonos, Yiannis (2020). EVDodgeNet: Deep Dynamic Obstacle Dodging with Event Cameras. In: 2020 IEEE International Conference on Robotics and Automation (ICRA), Paris, France, 1 July 2020 - 1 October 2020. IEEE, 10651-10657.

Abstract

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.

Abstract

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.

Statistics

Citations

Dimensions.ai Metrics

Altmetrics

Downloads

205 downloads since deposited on 17 Dec 2020
72 downloads since 12 months
Detailed statistics

Additional indexing

Item Type:Conference or Workshop Item (Paper), refereed, original work
Communities & Collections:03 Faculty of Economics > Department of Informatics
Dewey Decimal Classification:000 Computer science, knowledge & systems
Scopus Subject Areas:Physical Sciences > Software
Physical Sciences > Control and Systems Engineering
Physical Sciences > Artificial Intelligence
Physical Sciences > Electrical and Electronic Engineering
Scope:Discipline-based scholarship (basic research)
Language:English
Event End Date:1 October 2020
Deposited On:17 Dec 2020 09:07
Last Modified:06 Mar 2024 14:33
Publisher:IEEE
ISBN:978-1-7281-7395-5
OA Status:Green
Publisher DOI:https://doi.org/10.1109/icra40945.2020.9196877
Related URLs:https://ieeexplore.ieee.org/document/9196877
Other Identification Number:merlin-id:20312
  • Content: Accepted Version