Submission Deadline: March 15, 2020
Scope: Future applications of robotics and autonomous systems will involve increasingly large numbers of collaborative robots, sensors, and unmanned vehicles that are each capable of collecting, processing, and acting upon information with little or no human intervention. By sharing and coordinating information, plans, and decisions, these very-large-scale robotic (VLSR) networks can dramatically improve their performance in various industrial and military applications, including but not limited to space domain awareness, maritime surveillance, remote sensing, process monitoring and advanced manufacturing, and search and rescue. Sensing and control of collaborative agents, however, present many technical challenges, including the required computations that increase with the number of agents, and the challenge of accounting for information and uncertainties propagating through the network. It is well-known, for example, that the optimal control of N collaborative agents for path planning and obstacle avoidance is a PSPACE-hard problem. Also, while necessary for performing basic tasks such as localization, mapping, and object tracking, most sensing and estimation theories suffer from the challenge of large state-spaces and variable dimensionality.
One promising approach that has emerged in recent years is the use of probability density function (PDF) based methods and partial differential equation (PDE) models for deriving decentralized control strategies that scale up to VLSR networks comprised of hundreds of agents. At the same time, random finite set (RFS) theory and finite set statistics (FISST) have emerged as a unifying approach to estimation and tracking via multi-object PDFs that can be used to describe the state of multiple objects utilizing multiple sensor measurements.
We solicit original contributions for a special issue of theIEEE Transactions on Control of Network Systems (TCNS)which propose new scalable theory and algorithms for solving emerging challenges in the control and estimation of complex, large-scale systems. We seek rigorous contributions on new theoretical analysis and new methods for VLSR control, tracking, and estimation, as well as emerging VLSR applications of PDE and PDF-based methods, as well as other novel approaches.
The control of VLSR networks will require new methods for multi-object tracking and estimation, as well as scalable solutions to planning and decision making over multiple spatiotemporal scales. PDE models, such as the advection-diffusion equation, have been proven to be one useful tool for modeling spatiotemporal dynamics of VLSR networks. Using calculus of variations and PDE-constrained optimal control methods, PDE models have been shown useful for predicting and controlling VLSR behaviors, as well as biological, social, and economic systems.
However, this area of research requires more rigorous formalisms for relating discrete and continuum representations, and for providing large-scale control performance and robustness guarantees. Therefore, this special issue solicits original contributions that explore the use of RFS and other emerging theories on complex systems for representing the statistical properties of large collections of agents by compact and efficient representations, such as multi-object PDFs. Furthermore, probabilistic methods have been shown to generalize traditional Bayesian filtering and estimation to include flexible detection, sensing, and dynamic behaviors to many agents. However, several unsolved technical issues remain, including but not limited to the scalability of information value functions, the assimilation of heterogeneous data, and how to determine the object dynamics from data.
Topics: The theme of the special issue will be emerging theoretical and computational methods for the analysis, control, and optimization of VLSR networks.
Special Topics include but not limited to:
- Controllability and observability in PDE models of VLSR networks
- Optimality conditions for PDE and PDF-based control and estimation
- Efficient numerical and analytical solutions to large-scale control problems
- Solutions to large-scale control problems that utilize RFS theory
- New PDE or RFS models and state representations of VLSR systems
- New theory and analysis of VLSR discrete and continuous representations and dynamics
- Probabilistic solutions to swarming systems
- New theoretical results in stochastic control of large-scale systems leveraging RFS theory
- New formulations of estimation and control problem using RFS theory
- New applications of RFS theory to networked systems
All papers submitted to this special issue will be subject to peer review in accordance with the established practice of the IEEE Transactions on Control of Network Systems. Papers that do not fall within the scope of the special issue will be returned to the authors without review, to enable them to submit them as regular papers through the normal channels. The manuscript format should follow the guidelines posted on the TCNS web site. Please submit your paper through CONES, the TCNS on-line submission site.
Silvia Ferrari, Professor, Sibley School of Mechanical and Aerospace Engineering, Cornell University, Ithaca, NY (email)
Richard Linares, Charles Stark Draper Assistant Professor, Department of Aeronautics and Astronautics, Massachusetts Institute of Technology, Cambridge, MA (email)
Thomas A. Wettergren, Senior Technologist (ST) for Operational and Information Science, Naval Undersea Warfare Center (NUWC), Newport, RI (email)
Keith LeGrand, Senior Member of Technical Staff, Sandia National Laboratories, Albuquerque, NM (email)
|Submissions Open: November 11, 2019||Submission Deadline: March 15, 2020|
|First Decision: May 2020||Acceptance: October 2020|
|Final Manuscripts Due: December 2020||Publication: June 2021|
Submissions are accepted through the TCNS Submission site CONES. You must have an IEEE PIN number to submit a manuscript. In the CONES site, please select this special issue for your manuscript submission. To prepare your paper, please consult the TCNS Information for Authors.