21-BI-MAS: RSS 2021 Workshop on Behavioral Inference of Remotely Sensed Multi-agent Systems Virtual Virtual, Spain, July 12-17, 2021 |
Conference website | http://webdiis.unizar.es/~edumonti/21_RSS_WS/ |
Submission link | https://easychair.org/conferences/?conf=21bimas |
Submission deadline | July 2, 2021 |
Unsupervised learning of interaction dynamics and behavior from multi-agent observations has broad applications in physics, computer vision and robotics. Examples of multi-agent dynamic scenarios of interest in robotics include autonomous driving, livestock monitoring or competitive team games, such as the robocup. In all these problems, we can identify multiple agents (vehicles, animals, players, etc.) that dynamically interact with each other in complex ways that may be difficult to sense and analyze. Accurate modeling of their intended behavior and actions can improve not only a robot's capacity to perceive the world, as in SLAM, but also its capacity to make informed decisions and act accordingly.
A successful approach to these issues requires the use of advanced inference algorithms that utilize state-of-the-art machine learning, data analysis, and dynamical system modeling. In addition, specialized sensors, such as EO/IR or event-based cameras, may play an important role in the perception process by providing alternative modalities with contrasting and complementary capabilities that help discriminate characteristic actions. More importantly, the interplay between inference algorithms and sensors can prove critical in the development of successful robotic systems that are able to properly discern and identify multi-agent behaviors.
We encourage submissions of two types:
-
Short papers (4 pages in RSS conference format)
-
Novel ideas presenting preliminary results of ongoing research
-
Short versions of papers already published and/or presented at recent conferences associated with other research communities, such as machine learning (e.g., NIPS, AAAI), control (e.g., CDC, ACC, ECC), computer vision (e.g., ICCV, ECCV), etc.
-
-
Regular papers (8 pages in RSS conference format) for original, not previously published, contributions.
Submissions will be evaluated according to their quality and their interest in the context of the workshop. Accepted papers will be included in the workshop program with a short spotlight video followed by a Q&A interactive session.
Papers should be sent at https://easychair.org/conferences/?conf=21bimas
Topics of interest include, but are not limited to:
-
Multi-agent behavior representation and estimation
-
Multi-agent and swarm dynamics inference
-
Behavioral inference from multi-modal data
-
Active perception of multi-agent systems
-
Communication network inference
-
Emerging data analysis techniques, e.g. TDA
-
Pattern-of-life Discovery
-
Graph Neural Networks
-
Causal Representation Learning
-
SLAM in dynamic environments
-
Multi-agent Imitation Learning
-
Human-Robot Interaction Modelling
-
Multi-agent Reinforcement Learning
Important Dates
June 20th: Submission deadline (DEADLINE EXTENDED)
July 1st: Notification of acceptance
July 12-16: Workshop day (TBD)