Vision-Based Sense and Avoid Systems for Unmanned Aerial Vehicles

Three quadrotors used to perform our experiments.

This collaborative project aims to have fully autonomous Unmanned Aerial Vehicles (UAVs) for civil applications able to safely fly and share a common airspace without human supervision. More specifically, our work is on algorithms of the Sense and Avoid (SAA) system, fusing information from the sensors to track other UAVs and use those tracks to avoid collision if necessary. In this project, we focus on having a single camera as main sensor.

Our approach is to adapt the algorithms, both tracking and avoidance, to the limited field of view (FOV) of the camera. Because it is not possible to reliably track a target that is not the sensor’s FOV, the motion is constrained to guarantee collision-free trajectories.

To validate our approach, we run experiments an indoor facility equipped with a motion capture system, providing millimetric position accuracy. Our experiments are performed using small quadrotors that embed the sensing and computation necessary to perform the SAA.

Team and Collaborators

Sponsors and Research Period

This project was sponsored by Honeywell

Videos

Publications

2019

A Comparative Study of Collision Avoidance Algorithms for Unmanned Aerial Vehicles: Performance and Robustness to Noise

S. A. Roelofsen; D. Gillet; A. Martinoli 

2019. 13th International Symposium on Distributed Autonomous Robotic Systems (DARS), London, ENGLAND, Nov 07-09, 2016. p. 75-88. DOI : 10.1007/978-3-319-73008-0_6.

2017

Vision-Based Sense and Avoid Algorithms for Unmanned Aerial Vehicles

S. A. Roelofsen / A. Martinoli; D. Gillet (Dir.)  

Lausanne, EPFL, 2017. 

Collision Avoidance with Limited Field of View Sensing: A Velocity Obstacle Approach

S. A. Roelofsen; D. Gillet; A. Martinoli 

2017. IEEE International Conference on Robotics and Automation, Singapore, May 29 to June 3, 2017. p. 1922-1927. DOI : 10.1109/ICRA.2017.7989223.

2016

Environmental Field Estimation with Hybrid-Mobility Sensor Networks

W. C. Evans; D. da Cruz Baptista Dias; S. A. Roelofsen; A. Martinoli 

2016. IEEE International Conference on Robotics and Automation (ICRA), Stockholm, Sweden, p. 5301-5308. DOI : 10.1109/ICRA.2016.7487741.

3D Collision Avoidance Algorithm for Unmanned Aerial Vehicles with Limited Field of View Constraints

S. A. Roelofsen; A. Martinoli; D. Gillet 

2016. Conference on Decision and Control, Las Vegas, Nevada, USA, December 12-14, 2016. p. 2555-2560. DOI : 10.1109/CDC.2016.7798647.

Vision-Based Unmanned Aerial Vehicle Detection and Tracking for Sense and Avoid Systems

K. R. Sapkota; S. A. Roelofsen; A. Rozantsev; V. Lepetit; D. Gillet et al. 

2016. IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Daejeon, Korea, October 9-14, 2016. p. 1556-1561. DOI : 10.1109/IROS.2016.7759252.

2015

Reciprocal Collision Avoidance For Quadrotors Using On-board Visual Detection

S. A. Roelofsen; D. Gillet; A. Martinoli 

2015. International Conference on Intelligent Robots and Systems, Hamburg, Germany, September 28 – October 2, 2015. DOI : 10.1109/IROS.2015.7354053.

Distributed Deconfliction Algorithm for Unmanned Aerial Vehicles with Limited Range and Field of View Sensors

S. A. Roelofsen; A. Martinoli; D. Gillet 

2015. American Control Conference, Chicago, IL, USA, July 1-3, 2015. p. 4356-4361. DOI : 10.1109/ACC.2015.7172014.