(DURIP) REASONS- RESILIENT, ADAPTIVE, SCALABLE, AUTONOMY IN NETWORKED SWARMS
Abstract
Advances in computing, sensing, communications and algorithms are enabling the development of swarms of UAVs, UGVs and other mobile robots to autonomously coordinate on complex tasks such as surveillance, collective construction, search-and-rescue, situational awareness for ground troops, and exploration of hazardous environments. Several current projects exemplify these developments including the MBZIRC challenge, the DARPA Sub-terranean challenge and the DARPA OFFSET program. However, these efforts are aimed mainly at demonstrating the feasibility of robot swarms without answering fundamental questions about their scalability, efficiency, resilience and correctness of function. Our overarching vision towards addressing these fundamental questions involves the advancement and integration of the following core paradigms- 1) decentralized decision making in robot networks; 2) Resilience in Real-World Networks. Decentralized Decision Making in Robot Networks- Multirobot coordination requires several challenges to be solved in a decentralized manner including multirobot task allocation (MRTA), coverage planning, distributed state estimation and others. While there is a lot of work on MRTA, we have shown recently that realistic-lossy communication lose guarantees provided by such algorithms. We will evaluate more recent methods proposed recently, and evaluate the effect of realistic networking on such algorithms. Similarly, there is work in coverage path planning including our own on distributed algorithms to perform search. We will evaluate this work and others in realistic conditions as well as propose novel decentralized algorithms to suit our objectives. Resilience in Real-World Networks- It is common to model infrastructure networks, network of robots and networks embodying other cyber-physical systems (involving flow of information, energy, traffic etc.) as directed graphs. At a time where both natural hazards and cyber-attacks pose increasing threats to such complex network, it is important to be able to characterize the resilience of such networks. However, classical graph methods and associated heuristics fail to effectively capture the structure of networks jointly at the local and global level. Instead, we aim to explore the use of topological descriptors of graphs, which have been recently shown to be effective in capturing higher order interactions and structure. This will allow more effective resilience characterization, as well as generalization of such charcateristics across different networks of a given type. This proposal will bring together researchers in Computer Science, Mechanical and Aerospace, Industrial Systems and several labs in the School of Engineering at University at Buffalo to build a world-class testbed to study complex networks problems in multi-robot systems. This includes the Community of Excellence in Sustainable Manufacturing and Advanced Robotics Technologies (SMART CoE), DRONES Lab, LAIRS Lab, OPTIMATOR, ISTL, ADAMS Lab, and the AI institute. The proposed infrastructure will allow us to bridge the extensive autonomous vehicle infrastructure at UB with ground and aerial robotics infrastructure, state-of-the-art indoor and outdoor motion capture facility as well as modern sensors such as LiDARs and cameras that will enable these robots for fully autonomous function. Co-PIs have ongoing grants from DoD, NSF, NY-DoT and others that would greatly benefit from this infrastructure.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 07, 2023
- Source ID
- FA95502210169
Entities
People
- Karthik Dantu
Organizations
- Air Force Office of Scientific Research
- Research Foundation for the State University of New York
- United States Air Force