xSWARM 2020 - March 25 & 26th
Call for Participants
Identifying, modeling, characterizing, and predicting the control strategies and behaviors of partially observable swarms is a problem that exists in multiple disciplines across both natural and engineered systems. A particular challenge is in the case of “black box” swarms or—xSwarms— for which there is a lack of a priori information about the internal control and decision-making mechanisms and only partial information about the system’s perceptual and communication capabilities.
This is a problem that can occur in the study of natural collective systems, but also may increasingly become of importance to both military and first responder organizations as engineered swarming systems become easier to develop and field for a wide range of applications, including commercial ones. This generates a need for methods that that are appropriate for a range of situations including real-time or near real-time, and with varying degrees of automation and reliability. The significant theoretical, computational, experimental, and data challenges to progress are the focus of this workshop.
The goal of this workshop is to bring together researchers across diverse fields to provoke sharing of research, discussion, and brainstorming for the development of new tools for effective xSwarm engagement. Attendees will participate through presentations of relevant research or suggested areas of inquiry followed by question and answer sessions, or through panel discussions.
Topics are overlapping and include, but are not limited to:
- xSwarm Modeling – Standard swarm modeling approaches include bio-inspired models, centralized & decentralized flocking models, physics-based PDE’s, human crowd models, game theory implementations, and rule based algorithms. Can these be applied fruitfully to xSwarms? What other approaches might be used?
- Classification & Identification – How can we learn underlying strategies through observation and interaction? Are there invariant features of a particular swarm that can be used for classification (such as communication structure, obstacle avoidant reactivity, or a reachability metric)? Relevant tools might include machine learning, pattern recognition, or observability analysis
- Prediction and Control – How might we apply existing theory in observability and control? How can we approach modeling & classification in ways with the most utility for prediction and control? What displays or decision tools might be used to facilitate this? What are tradeoffs to consider between interacting for identification vs control?
We invite all researchers with interest in this topic to submit a brief abstract proposing a presentation topic or proposing a panel. Presentation slots are 20 minutes, not including the question and answer session which will follow. Panel abstracts should propose a question of interest and submitters will lead the panel. Other panel participants can be optionally suggested, otherwise the panel will be populated by the organizers with other participants. Estimated panel lengths are 30-45 minutes.
For the talks as well as the panels this workshop encourages open ended questions as subject matter. What topics are missing from the conversation? What hasn’t been developed yet? What are some dots you’d like to see connected?
Isaac Kaminer (email@example.com)
Wei Kang (firstname.lastname@example.org)
Brian Bingham (email@example.com)
Claire Walton (firstname.lastname@example.org)
Sean Kragelund (email@example.com)
- Registration opens:
- November 15, 2019
- Deadline for abstract submission:
- January 1, 2020
- Schedule announced:
- March 8, 2020
- March 25 & 26, 2020