Niranjan Suri
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Niranjan Suri has a total of 54 co-authors. Here are the top co-authors by number of shared publications:
- Brian Jalaian
Department of Robotics and Intelligent Systems, University of West Florida, Pensacola, FL, USA
2 shared publications - Cesare Stefanelli
University of Ferrara, Ferrara, Italy
2 shared publications - Filippo Poltronieri
Department of Engineering, University of Ferrara, Ferrara, Italy
2 shared publications - Jeffrey M. Bradshaw
Florida Institute for Human and Machine Cognition (IHMC) Pensacola Florida USA
2 shared publications - Mauro Tortonesi
University of Ferrara, Ferrara, Italy
2 shared publications - Alessandro Amato
Institute for Human and Machine Cognition (IHMC) Pensacola Florida USA
1 shared publication
Latest Publications
- Open Access
- 8 November 2026
- Published by Wiley in AI Magazine
- Vol. 47 (4)
Safety in numbers: Tackling the deliberative, performative, and infrastructural roots of multi‐agent misalignment
Multi‐agent systems (MAS)—a uniquely important subset of the wider field commonly called “Agentic AI”—enable complex work processes involving distributed networks of multiple AIs, people, and technology tools. However, they also carry big risks. Root causes of catastrophic safety failures in multi‐agent collaboration (as in human collaboration) can be classed into three overlapping categories: 1. Deliberative —deviating from shared MAS commitments to fulfill explicitly expressed joint intentions; 2. Performative —inability to coordinate well or to adapt to changing circumstances; and 3. Infrastructural —safety, reliability, and trust in the operating environment. In the spirit of the international Multi‐Agent System Safety and Teamwork (MASST) initiative (Bradshaw et al., 2026), our multi‐disciplinary research team seeks to tackle the root causes of these three challenges through development, experimentation, and deployment of an integrated multi‐layered safety pipeline, enabling “safety in numbers” as increasingly powerful multi‐agent systems are deployed. Although this proposal is merely an initial sketch, it is our hope that sharing these ideas will inspire others to contribute to the discussion as we continue moving forward. We look forward not only to exciting discoveries within laboratory settings while also being committed to improving safe rollout of large‐scale MAS into the world to perform work in high‐consequence applications.
- 8 November 2026
- Published by Wiley in AI Magazine
- Vol. 47 (4)
Open AccessSafety in numbers: Tackling the deliberative, performative, and infrastructural roots of multi‐agent misalignment
Multi‐agent systems (MAS)—a uniquely important subset of the wider field commonly called “Agentic AI”—enable complex work processes involving distributed networks of multiple AIs, people, and technology tools. However, they also carry big risks. Root causes of catastrophic safety failures in multi‐agent collaboration (as in human collaboration) can be classed into three overlapping categories: 1. Deliberative —deviating from shared MAS commitments to fulfill explicitly expressed joint intentions; 2. Performative —inability to coordinate well or to adapt to changing circumstances; and 3. Infrastructural —safety, reliability, and trust in the operating environment. In the spirit of the international Multi‐Agent System Safety and Teamwork (MASST) initiative (Bradshaw et al., 2026), our multi‐disciplinary research team seeks to tackle the root causes of these three challenges through development, experimentation, and deployment of an integrated multi‐layered safety pipeline, enabling “safety in numbers” as increasingly powerful multi‐agent systems are deployed. Although this proposal is merely an initial sketch, it is our hope that sharing these ideas will inspire others to contribute to the discussion as we continue moving forward. We look forward not only to exciting discoveries within laboratory settings while also being committed to improving safe rollout of large‐scale MAS into the world to perform work in high‐consequence applications.
- Open Access
- 9 September 2025
- Published by Elsevier in Journal of Network and Computer Applications
- Vol. 243
RoamML distributed continual learning: Adaptive and flexible data-driven response for disaster recovery operations
- 9 September 2025
- Published by Elsevier in Journal of Network and Computer Applications
- Vol. 243
Open AccessRoamML distributed continual learning: Adaptive and flexible data-driven response for disaster recovery operations
- 30 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 219-224
IoT in Coalition Federated Operations: Multi-National C2 Integration and Technical Interoperability Experiments
The Internet of Things (IoT) is an emerging and disruptive technology in the military domain. Military technology innovators can exploit the IoT world to promote information superiority in the battlespace through novel and abundant sensory capabilities. Our research was conducted as part of the NATO research task group IST-176 on "Federated Interoperability of Military C2 and IoT Systems" to investigate how IoT data can be ingested into C2 systems in a federated coalition operation. Our approach leverages industrial standard protocols and formats, while including optional fields to convey military-specific unit information along with IoT sensor data. Our experiments demonstrate and validate the approach in which systems from different nations are connected in a federated environment, delivering a high success rate in message delivery and enabling a consistent display of the situational picture across different C2 systems. The results of the experiments demonstrate the feasibility of integrating IoT information in a military information flow, to support collective C2.
- 30 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 219-224
IoT in Coalition Federated Operations: Multi-National C2 Integration and Technical Interoperability Experiments
The Internet of Things (IoT) is an emerging and disruptive technology in the military domain. Military technology innovators can exploit the IoT world to promote information superiority in the battlespace through novel and abundant sensory capabilities. Our research was conducted as part of the NATO research task group IST-176 on "Federated Interoperability of Military C2 and IoT Systems" to investigate how IoT data can be ingested into C2 systems in a federated coalition operation. Our approach leverages industrial standard protocols and formats, while including optional fields to convey military-specific unit information along with IoT sensor data. Our experiments demonstrate and validate the approach in which systems from different nations are connected in a federated environment, delivering a high success rate in message delivery and enabling a consistent display of the situational picture across different C2 systems. The results of the experiments demonstrate the feasibility of integrating IoT information in a military information flow, to support collective C2.
- Open Access
- 28 April 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-6
Improving Object Detection Robustness against Natural Perturbations through Synthetic Data Augmentation
Robustness against real-world distribution shifts is crucial for the successful deployment of object detection models in practical applications. In this paper, we address the problem of assessing and enhancing the robustness of object detection models against natural perturbations, such as varying lighting conditions, blur, and brightness. We analyze four state-of-the-art deep neural network models, Detr-ResNet-101, Detr-ResNet-50, YOLOv4, and YOLOv4-tiny, using the COCO 2017 dataset and ExDark dataset. By simulating synthetic perturbations with the AugLy package, we systematically explore the optimal level of synthetic perturbation required to improve the models’ robustness through data augmentation techniques. Our comprehensive ablation study meticulously evaluates the impact of synthetic perturbations on object detection models’ performance against real-world distribution shifts, establishing a tangible connection between synthetic augmentation and real-world robustness. Our findings not only substantiate the effectiveness of synthetic perturbations in improving model robustness, but also provide valuable insights for researchers and practitioners in developing more robust and reliable object detection models tailored for real-world applications.
- 28 April 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-6
Open AccessImproving Object Detection Robustness against Natural Perturbations through Synthetic Data Augmentation
Robustness against real-world distribution shifts is crucial for the successful deployment of object detection models in practical applications. In this paper, we address the problem of assessing and enhancing the robustness of object detection models against natural perturbations, such as varying lighting conditions, blur, and brightness. We analyze four state-of-the-art deep neural network models, Detr-ResNet-101, Detr-ResNet-50, YOLOv4, and YOLOv4-tiny, using the COCO 2017 dataset and ExDark dataset. By simulating synthetic perturbations with the AugLy package, we systematically explore the optimal level of synthetic perturbation required to improve the models’ robustness through data augmentation techniques. Our comprehensive ablation study meticulously evaluates the impact of synthetic perturbations on object detection models’ performance against real-world distribution shifts, establishing a tangible connection between synthetic augmentation and real-world robustness. Our findings not only substantiate the effectiveness of synthetic perturbations in improving model robustness, but also provide valuable insights for researchers and practitioners in developing more robust and reliable object detection models tailored for real-world applications.
- 4 May 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-8
Fountain Coding for Information Protection in Tactical Networks
Tactical edge networks are operated in harsh conditions where ensuring information protection is a challenging issue. An insider threat is one of the major issues that arises in such scenarios. For instance, an attacker can compromise one of the nodes and intercept information that is being relayed to other nodes. In this paper, we propose Fountain Coding for Information Protection (FCIP) to mitigate this threat. The source node encodes information from k source symbols into m messages before sending them over the network and the receiver must recover at least k messages to decipher the information. To mitigate the insider threat, we route the encoded packets along disjoint paths. However, this incurs a trade-off with information availability as the encoded packets are routed along sub-optimal paths. It is also necessary to determine the portion of traffic to encode as encoding all the traffic is resource intensive. We formulate a multi-objective optimization problem that takes into account information protection while trading-off with information availability and considers bandwidth, energy and availability constraints. We model a risk metric to determine the cost of information protection that is provided and a cost metric to quantify the cost of information availability. We demonstrate the benefit of FCIP in our performance study by comparing it against a shortest path routing solution and provide optimal results for FCIP. We analyze the trade-off between information protection and availability and provide insight on the performance and benefits of FCIP for different operational scenarios.
- 4 May 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-8
Fountain Coding for Information Protection in Tactical Networks
Tactical edge networks are operated in harsh conditions where ensuring information protection is a challenging issue. An insider threat is one of the major issues that arises in such scenarios. For instance, an attacker can compromise one of the nodes and intercept information that is being relayed to other nodes. In this paper, we propose Fountain Coding for Information Protection (FCIP) to mitigate this threat. The source node encodes information from k source symbols into m messages before sending them over the network and the receiver must recover at least k messages to decipher the information. To mitigate the insider threat, we route the encoded packets along disjoint paths. However, this incurs a trade-off with information availability as the encoded packets are routed along sub-optimal paths. It is also necessary to determine the portion of traffic to encode as encoding all the traffic is resource intensive. We formulate a multi-objective optimization problem that takes into account information protection while trading-off with information availability and considers bandwidth, energy and availability constraints. We model a risk metric to determine the cost of information protection that is provided and a cost metric to quantify the cost of information availability. We demonstrate the benefit of FCIP in our performance study by comparing it against a shortest path routing solution and provide optimal results for FCIP. We analyze the trade-off between information protection and availability and provide insight on the performance and benefits of FCIP for different operational scenarios.
- Open Access
- 1 November 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 89 90, 809-814
MARGOT: Dynamic IoT Resource Discovery for HADR Environments
Smart City services leverage sophisticated IT architectures whose assets are deployed in dynamic and heterogeneous computing and communication scenarios. Those services are particularly interesting for Humanitarian Assistance and Disaster Relief (HADR) operations in urban environments, which could improve Situation Awareness by exploiting the Smart City IT infrastructure. To this end, an enabling requirement is the discovery of the available Internet-of-Things (IoT) resources, including sensors, actuators, services, and computing resources, based on a variety of criteria, such as geographical location, proximity, type of device, type of capability, coverage, resource availability, and communication topology / quality of network links. To date, no single standard has emerged that has been widely adopted to solve the discovery challenge. Instead, a variety of different standards have been proposed and cities have either adopted one that is convenient or reinvented a new standard just for themselves. Therefore, enabling discovery across different standards and administrative domains is a fundamental requirement to enable HADR operations in Smart Cities. To address these challenges, we developed MARGOT (Multi-domain Asynchronous Gateway Of Things), a comprehensive solution for resource discovery in Smart City environments that implements a distributed and federated architecture and supports a wide range of discovery protocols.
- 1 November 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 89 90, 809-814
Open AccessMARGOT: Dynamic IoT Resource Discovery for HADR Environments
Smart City services leverage sophisticated IT architectures whose assets are deployed in dynamic and heterogeneous computing and communication scenarios. Those services are particularly interesting for Humanitarian Assistance and Disaster Relief (HADR) operations in urban environments, which could improve Situation Awareness by exploiting the Smart City IT infrastructure. To this end, an enabling requirement is the discovery of the available Internet-of-Things (IoT) resources, including sensors, actuators, services, and computing resources, based on a variety of criteria, such as geographical location, proximity, type of device, type of capability, coverage, resource availability, and communication topology / quality of network links. To date, no single standard has emerged that has been widely adopted to solve the discovery challenge. Instead, a variety of different standards have been proposed and cities have either adopted one that is convenient or reinvented a new standard just for themselves. Therefore, enabling discovery across different standards and administrative domains is a fundamental requirement to enable HADR operations in Smart Cities. To address these challenges, we developed MARGOT (Multi-domain Asynchronous Gateway Of Things), a comprehensive solution for resource discovery in Smart City environments that implements a distributed and federated architecture and supports a wide range of discovery protocols.
- 3 December 2010
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Systems, Man and Cybernetics, Part C (Applications and Reviews)
- Vol. 41 (3) , 365-375
Automated Conflict Resolution Utilizing Probability Collectives Optimizer
Rising manned air traffic and deployment of unmanned aerial vehicles in complex operations requires integration of innovative and autonomous conflict detection and resolution methods. In this paper, the task of conflict detection and resolution is defined as an optimization problem searching for a heading control for cooperating airplanes using communication. For the optimization task, an objective function integrates both collision penalties and efficiency criteria considering airplanes' objectives (waypoints). The probability collectives optimizer is used as a solver for the specified optimization task. This paper provides two different implementation approaches to the presented optimization-based collision avoidance: 1) a parallel computation using multiagent deployment among participating airplanes and 2) semicentralized computation using the process-integrated-mechanism architecture. Both implementations of the proposed algorithm were implemented and evaluated in a multiagent airspace test bed AGENTFLY. The quality of the solution is compared with a negotiation-based cooperative collision avoidance method - an iterative peer-to-peer algorithm.
- 3 December 2010
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Systems, Man and Cybernetics, Part C (Applications and Reviews)
- Vol. 41 (3) , 365-375
Automated Conflict Resolution Utilizing Probability Collectives Optimizer
Rising manned air traffic and deployment of unmanned aerial vehicles in complex operations requires integration of innovative and autonomous conflict detection and resolution methods. In this paper, the task of conflict detection and resolution is defined as an optimization problem searching for a heading control for cooperating airplanes using communication. For the optimization task, an objective function integrates both collision penalties and efficiency criteria considering airplanes' objectives (waypoints). The probability collectives optimizer is used as a solver for the specified optimization task. This paper provides two different implementation approaches to the presented optimization-based collision avoidance: 1) a parallel computation using multiagent deployment among participating airplanes and 2) semicentralized computation using the process-integrated-mechanism architecture. Both implementations of the proposed algorithm were implemented and evaluated in a multiagent airspace test bed AGENTFLY. The quality of the solution is compared with a negotiation-based cooperative collision avoidance method - an iterative peer-to-peer algorithm.
- 1 June 2008
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 199-202
Policy Management across Multiple Platforms and Application Domains
One of the challenges of building a policy management framework is making it flexible enough to handle differences in both policy semantics and enforcement strategies across multiple platforms and application domains. The system must be expressive enough in each application domain to provide the richness needed for interesting policies. It must also provide a simple and flexible enforcement mechanism for adaptation to a variety of systems. In this paper we discuss the application of the KAoS policy services framework to human-robot teamwork - an application that involves a variety of application domains and enforcement at different levels of control; from low level network resource control to high level organizational constraints and coordination management. The study culminated in an outdoor field exercise that required coordination of mixed sub teams composed of two people and five robots whose task was to find and apprehend an intruder on a Navy pier.
- 1 June 2008
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 199-202
Policy Management across Multiple Platforms and Application Domains
One of the challenges of building a policy management framework is making it flexible enough to handle differences in both policy semantics and enforcement strategies across multiple platforms and application domains. The system must be expressive enough in each application domain to provide the richness needed for interesting policies. It must also provide a simple and flexible enforcement mechanism for adaptation to a variety of systems. In this paper we discuss the application of the KAoS policy services framework to human-robot teamwork - an application that involves a variety of application domains and enforcement at different levels of control; from low level network resource control to high level organizational constraints and coordination management. The study culminated in an outdoor field exercise that required coordination of mixed sub teams composed of two people and five robots whose task was to find and apprehend an intruder on a Navy pier.