Abstract
I present my results from a “winter of simulation” in which fieldwork was largely impractical, prompting me to use virtual experiments to de-risk ideas that span above, on, and under water. I introduce SailSwarmSwIM, a scalable simulator for swarms of autonomous sailboats in which wind is treated as a first-class system state, and classical collective-motion models must be adapted to sailing constraints, such as no-go zones and tacking. The key result is that gusts and waves can drive repeatable, non-monotonic changes in swarm metrics (area, polarization), highlighting that environment fields can shape collective behavior as strongly as interaction rules. Underwater, I present BIND-USBL, a HoloOcean-based study of heterogeneous ASV–AUV teams in GPS-denied conditions, where a surface vehicle acts as a mobile base station and provides event-triggered USBL corrections to bound IMU drift under a limited acoustic servicing budget.
About the speaker
Hello! I'm Pranav Kedia, a Ph.D. researcher at the cyber physical group supervised by Prof. Dr. Heiko Hamann at the University of Konstanz, where I'm affiliated with the Centre for the Advanced Study of Collective Behaviour. I am working on a swarm of autonomous sailboats optimized for long-duration endurance. This research has been a combination of learning about sailing and applying my skills and knowledge in Embedded systems and Robotics. Previously, I worked on Bee Waggle Robot in the EU H2020 project 'Hiveopolis' at the Biorobtics Lab led by Prof. Dr. Tim Landgraf at the Free University of Berlin. I completed my master's majoring in electronics and communication with a specialization in embedded systems at IIIT-B. Previously, I was fortunate to be part of the Surgical and Assistive Robotics Lab, led by Prof. Madhav Rao. I was also part of the ARMS lab, led by Dr. Arpita Sinha at IIT Bombay, India, in the summer of 2019. Prof. Sachit Rao had supervised me on guided coursework and non-course projects. I enjoy working on Bio-inspired Swarm Robotics and Swarm Intelligence methods.
