If you work on such topics and you have high-quality original research 
results, you are welcomed to send us a paper title, authors and 
1-paragraph (1/2 page) abstract.

Abstracts will be peerly reviewed by third persons. The best 6 of them 
will be included in the special session proposal. If the special 
session proposal is accepted, any relevant paper (certainly the ones 
included in the proposal plus more) will be invited through an open 
paper call (to come out then) to be submitted as a regular IROS 
special session paper and will be peerly reviewed as such.

Right now, we only need a paper title, author list (with affiliations) 
and abstract, to be sent to us by email (to the address 
imademlis@csd.auth.gr).

Please feel free to forward this to your students, as well as to other 
qualified labs you are in contact with.

We would greatly appreciate a fast contribution (till January 15, 2022).

Best regards,
Ioannis Pitas

IROS2022 Special Session  proposal TITLE: Robotics and AI for 
Autonomous UAV Cinematography

DESCRIPTION
The use of Unmanned Aerial Vehicles (UAVs, or “drones”) for 
professional media production is an existing commercial reality. 
Currently available consumer UAVs equipped with high-end cameras 
already exhibit a degree of cognitive and physical autonomy, relying 
on Artificial Intelligence (AI) and robotics technologies. Current 
research promises to further exploit the potential of autonomous 
functionalities in the immediate future, joining together various 
scientific fields that range from machine learning to computer 
graphics, perception and aerial robotics. The end-result is expected 
to yield portable flying robotic cameras with advanced intelligence 
concerning subject detection/tracking, cinematic shot execution, 
cinematography planning, 3D localization and environmental mapping, as 
well as autonomous obstacle avoidance combined with on-line motion 
re-planning. Thus, the exciting emerging area of highly autonomous 
UAVs for media production is gradually coalescing under the common 
theme of drone intelligence oriented towards capturing aesthetically 
pleasing footage in dynamic environments.
This Special Session will consider current research in this 
interdisciplinary subject, bringing together researchers working on 
deep learning, computer vision, computer graphics and aerial robotics. 
The goal will be to foster tighter synergies between complementary 
fields. Topics of interest may include, among others:

– Embedded computer vision for UAV cinematography
– Imitation and/or deep reinforcement learning for autonomous UAV 
cinematography
– Vision-based UAV localization and mapping
– UAV control methods for cinematic shot execution
– Mission/motion planning for autonomous UAV cinematography
– Multi-agent methods for autonomous UAV cinematography
– Computational aesthetics for autonomous UAV cinematography
– Autonomous shot selection for UAV cinematography
– Human-robot interfaces for UAV media production

NOVELTY AND MOTIVATION
This Special Session concerns a very timely topic with high industrial 
potential, given that fully/semi- automated drones are slowly emerging 
as a viable alternative to manually teleoperated ones, thanks to 
recent advances in robotics and AI. The variety of the underlying 
functionalities that enable such autonomy make this a highly 
interdisciplinary topic of exceptional interest to many types of 
experts: aerial robotics or AI researchers, computer graphics 
professionals, media technology providers, etc.

ORGANIZERS
Dr. Ioannis Mademlis (Aristotle University of Thessaloniki, Greece)
Prof. Ioannis Pitas (Aristotle University of Thessaloniki, Greece)
Prof. Rita Cunha (University of Lisbon, Portugal)