Abstract
This paper presents a system for autonomous cinematography with multiple drones. Drones are becoming a trend for aerial cinematography. The price for buying a commercial platform is decreasing every year, while their quality increases. Drones allow for new shots and perspectives, and they can be automated. Despite the extended use of drones, there are still challenges for teams of multiple drones which cooperate for autonomous cinematography. The proposed system tries to face these challenges, focusing on the actual execution of aerial shots. A set of canonical shots with specific parameters for autonomous implementation is compiled. This system includes a distributed scheduler to synchronize shots using an event-based mechanism, and an autonomous controller to provide smooth movements in both the drone and the camera, so the drones can take aesthetic shots. Moreover, the system considers safety in two levels: collision avoidance in the controller and an emergency management component to handle high level alarms (e.g. low battery)
| Original language | English |
|---|---|
| Title of host publication | Workshop on Signal Processing Computer vision and Deep Learning for Autonomous Systems, EUSIPCO2019 |
| Publisher | IEEE Signal Processing Society (SPS) |
| Publication status | Published - 2019 |
| Event | EUSIPCO 2019 Satellite Workshop on Signal Processing, Computer Vision and Deep Learning for Autonomous Systems - A Coruña, Spain Duration: 6 Sept 2019 → 6 Sept 2019 |
Workshop
| Workshop | EUSIPCO 2019 Satellite Workshop on Signal Processing, Computer Vision and Deep Learning for Autonomous Systems |
|---|---|
| Country/Territory | Spain |
| City | A Coruña |
| Period | 6/09/19 → 6/09/19 |