January 2024 – Present
This project aims to tackle the inefficiencies of current soft target security systems such as high costs and limited adaptability to new and emerging threats. These inefficiencies overwhelm human operators, resulting in delayed incident response times. The project proposes the creation of a new real-time semi-autonomous system that uses drones and ground robots equipped with various sensors to identify, report, and as a result, reduce threats in crowded areas with minimal human involvement. The focus is on learning about development, testing, and experimental evaluation techniques to improve the system for deploying teams of moving robots and drones capable of navigating different environments for better risk assessment and response to evolving threats.
The project utilizes a combination of real-time multi-modal sensor fusion, multi-agent coordination, and algorithms for unmanned systems. The sensors include mobile mm-wave radars, optical cameras, and inertial measuring units, while the agents are coordinated across multiple temporal and spatial scales to protect soft targets and crowded places. The project is currently testing the proposed framework at the Kostas Research Institute to evaluate the technology’s capability for enhancing situational awareness and mitigation strategies in real-world environments. These tests will be used to improve the real-time semi-autonomous system for a seamless transition at the project’s end.
This project advances the state of the art by integrating real-time multi-sensor fusion, enhancing decentralized robotic swarm understanding, and optimizing coordination of unmanned systems. Unlike current security systems, this project employs Radar Inertial Odometry (RIO), which fuses an mm-wave radar and an inertial measuring unit to provide Simultaneous Localization and Mapping (SLAM), enabling an autonomous robot to navigate in cities where GPS is degraded or denied after an attack. Additionally, this project aims to address the research gap related to coordinating multi-agent sensing, decision-making, and mitigation actions over an extended time by utilizing a decentralized approach for multi-scale (temporal and spatial) coordination in robotic autonomous systems.
In 2024, the project has focused on development, testing, and evaluation in three main technical areas: real-time multi-modal sensor fusion for robust perception, agent-assisted mitigation strategies, and multi-scale coordination of tasks and agents. These efforts lay the groundwork for testing at the Kostas Research Institute testbed.
This project is collaborating with MatrixSpace Inc., a member of SENTRY’s Industry Advisory Board, to make use of their multimodal sensor payloads. Furthermore, this project is working in collaboration with other SENTRY researchers to exchange datasets and experiment with innovative sensors. The project results will also be used to validate policies and procedures in simulators and experimental settings for continued research. These collaborations ensure the project’s solutions are practical, effective, and ready for real-world deployment.