The border between two regions is a barrier that intruders often try to cross. Detection of such intrusions require effective monitoring of  the barrier region by strategically deployed sensors. In our work we develop strategies and algorithms for effective barrier coverage using hovering UAVs equipped with downward facing vision sensors.

Amit Kumar and Debasish Ghose. "Barrier Coverage of a Belt with Varying Resolution Requirements Using UAVs," AIAA 2024-1174. AIAA SCITECH Forum. Florida, January 2024. 

Amit Kumar and Debasish Ghose, "Enhancing Resolution and Fault Tolerance of Barrier Coverage with Unmanned Aerial Vehicles", AIAA Journal of Aerospace Information Systems, 2024 (Accepted).

Using drones to monitor and maintain large 3-D structures like bridges, buildings, and large ships is challenging as the structures may not conform to regular shapes. In our work, we devise algorithms that allow effective monitoring of such structures using multiple autonomous drones. We use concepts from Lissajous curves to Fourier transforms to design trajectories that guarantee performance in specific coverage metrics. 

S. Nath, M. Baishya, and D. Ghose, Decentralised Coverage of a Large Structure Using Flocking of Autonomous Agents Having a Dynamic Hierarchy Model, Autonomous Robots, Vol. 46, pp. 617-643, 2022.

Suryadeep Nath and Debasish Ghose. "Dynamic Aerial Coverage of Stationary and Moving Structures Using Lissajous Curves," AIAA 2024-1991. AIAA SCITECH Forum. Florida, January 2024. 

When an automated emergency vehicle, such as an ambulance, needs to find its way through humans or human-operated vehicles, it has to deal with the motion dynamics of entities that can differ widely based on various human traits. Further, these entities may display different levels of cooperation with the emergency vehicles. Our objective is to develop AI/ML techniques to help the vehicle negotiate through such crowded environments by ascertaining the cooperativeness of the human entities.

Cislunar space is the three-dimensional region beyond Earth’s geosynchronous orbit but within the gravitational influence of the Earth and/or the Moon. Interesting new class of orbits emerge in the cislunar space that can be exploited for various missions such as interplanetary missions, earth and space body observation, etc. Our work is presently focused on the logistics of placing observation stations in cislunar orbits to observe other space assets in nearby cislunar orbits.

We introduce the Robot Autonomous Motion (RoAM) video dataset, which is collected with a custom-made turtlebot3 Burger robot in a variety of indoor environments recording various human motions from the robot’s ego-vision. The dataset also includes synchronized records of the LiDAR scan and all control actions taken by the robot as it navigates around static and moving human agents. The unique dataset provides an opportunity to develop and benchmark new visual prediction frameworks that can predict future image frames based on the action taken by the recording agent in partially observable scenarios or cases where the imaging sensor is mounted on a moving platform. 

Sarkar, M., Honkote, V., Das, D. and Ghose, D., 2023, August. Action-conditioned Deep Visual Prediction with RoAM, a new Indoor Human Motion Dataset for Autonomous Robots. In 2023 32nd IEEE International Conference on Robot and Human Interactive Communication (RO-MAN) (pp. 1115-1120). IEEE. 

Long-term video generation and prediction remain challenging tasks in computer vision, particularly in partially observable scenarios where cameras are mounted on moving platforms. The interaction between observed image frames and the motion of the recording agent introduces additional complexities. To address these issues, we introduce the Action-Conditioned Video Generation (ACVG) framework, a novel approach that investigates the relationship between actions and generated image frames through a deep dual Generator-Actor architecture. 

M. Sarkar and D. Ghose, "Action-conditioned video data improves predictability" accepted in WiCV, CVPR, Seattle, Washington, June 2024.

Integrating Unmanned Aerial Vehicles (UAVs) into airspace requires a reliable framework which is robust and scalable. CORRIDRONE is one such architecture that generates corridors for point-to-point traversal of drones. This work presents details about its central features like adaptive geofencing, drone compliance levels, and corridor geometry. Adaptive geo-fencing guarantees vehicle safety when multiple vehicles of different hardware and software capabilities are sharing the same airspace. Compliance levels, defined based on these capabilities of the UAVs, are an essential measure for determining geo-fence bounds. The lane geometry of CORRIDRONE is designed considering the aerodynamic aspects like downwash, which ensures in-flight stability of the UAVs. These features confirm safe transit of the vehicles and also ensures efficient operation of the system. The flexibility to accommodate multiple UAVs of varying compliance levels highlight the robustness of the proposed framework. 

Tony, L.A., Ratnoo, A. and Ghose, D., 2021, June. Lane geometry, compliance levels, and adaptive geo-fencing in CORRIDRONE architecture for urban mobility. In 2021 International Conference on Unmanned Aircraft Systems (ICUAS) (pp. 1611-1617). IEEE. 

Unmanned aerial system (UAS) traffic management of airspace is a domain that demands strategic management of unmanned aerial vehicles (UAVs) for smooth and conflict-free movement in the uncharted low-altitude G airspace. In the context of the previously proposed CORRIDRONE structure, UAV traffic has to be organized in a shared volume of airspace connecting two or more corridors for a network of multilane corridors, forming aerial intersections. This work proposes an intersection planning algorithm to provide no-conflict paths to the UAVs inside the intersection volume. Paths are modeled as a function of the lanes involved in the transition, and conflict resolution is achieved by changing lanes. Optimized solutions are found among the conflicted UAV paths, such that only a few paths need modifying, optimizing the number of lane changes and time spent in the intersection.

Nagrare, S.R., Ratnoo, A. and Ghose, D., Intersection planning for multilane unmanned aerial vehicle traffic management. AIAA Journal of Aerospace Information Systems, 21(3), pp.216-233, 2024. 

Animals, birds, and flies hunt their prey using different hunting behaviors. These hunting behaviors motivated some of the guidance laws, such as the pursuit strategy used in defense systems, unmanned aerial vehicles, and robot guidance problems. Hawks have been known to display a hunting strategy that instinctively makes them speed up whenever the prey moves in a straight line and slow down when the prey changes direction or maneuvers. In this work, we propose a new guidance strategy based on this behavior and show how it conserves energy for the interceptor.

Paul, N. and Ghose, D., Longitudinal-Acceleration-Based Guidance Law for Maneuvering Targets Inspired by Hawk’s Attack Strategy, AIAA Journal of Guidance, Control, and Dynamics, 46(7), pp. 1437-1447, 2023.