Research Questions: How can we coordinate connected robot fleets that are cooperating with a diverse array of stakeholders to achieve their group objectives? Can we scale the autonomy with the fleet size and geographic span with minimal re-configurations in long-term?
I research along game theory and deep multi-agent reinforcement learning to extend the fleet autonomy with long-term scalability and cooperative decision-making, focusing on the inter- and intra-fleet interactions.
- (2022 August) I presented our work on graphical game-theoretic swarm control at IEEE RO-MAN (International Conference on Robot and Human Interactive Communication (RO-MAN).
- (2022 June) Presented my research on graphical game -theoretic control at University of Sydney.
- (2022 May) I presented at the Workshop on Intelligent Aerial Robotics , International Conference on Robotics and Automation – 2022, Philadelphia.
- (2022 March) A new paper accepted to IEEE Robotics and Automation Letters (RA-L): “CoCo Games: Graphical Game-Theoretic Swarm Control for Communication-Aware Coverage” [Paper].
- (2022 February) The code base for ICRA-2021 paper, “Online Flocking Control of UAVs with Mean-Field Approximation” is released. Check it out: github.com/malintha/mean_field_flocking.
Recent WorkVideo for my presentation at 2022, IEEE Ro-Man, Napoli, Italy. You can download the slides for the presentation here. URL for the project website: https://malintha.github.io/coco/.
Online Flocking Control of UAVs with Mean-Field ApproximationIn this work we cast the aerial swarm flocking as a probabilistic inference and utilize mean-field approximation to infer the control actions for the robots in real time. This allows achieving dynamically feasible real-time flocking control with convergence guarantees borrowed from graphical games.
Flocking of Multi Aerial Vehicles Generation of flocking behavior for UAV swarms in confined areas.
MavSwarm A robust and light weight multi UAV dynamics simulator based on Robot Operating System (ROS). Find the code at https://github.com/Malintha/multi_uav_simulator
Formation Control and Navigation of a Quadrotor Swarm A swarm of Crazyflie nano drones are controlled by using rigid body dynamics. Published at International Conference on Unmanned Aircraft Systems (2019).
Aggressive and Precise Maneuvering of a Crazyflie Nano Drone This work includes generating smooth, piecewise trajectories with quadratic programming.