Sensing abstract behavioural patterns to allow coordinated fast response to disruptions in multi-robot systems

Standard grant project in years 2026-2028 supported by the Czech Science Foundation (GAČR) under research project No. 26-22419S

Applicant: Martin Saska

Main Goal:

To design an onboard perception and swarming framework that enables fully distributed, fast responses of cooperative and non-cooperative UAV teams to disruptions, exceeding the speed and acceleration of natural flocks. To gain insights into biological swarming processes from a robotic perspective.

Abstract:

This project aims to rigorously analyze the dynamics of large flocks of fast-flying birds, capturing both body and wing motion with high spatial and temporal resolution, and transfer these insights into AI-driven robotic swarms. We will investigate the motion features underlying abstract behavioral patterns that enable rapid, coordinated responses to disruptions.

Our study will consider negative disruptions (e.g., predator evasion; individual failures; uncertainties; imminent collisions) and positive ones (e.g., food source observation; cohesion with alignment for safety), to which animal groups react quickly in a fully distributed manner. Analogically, in aerial multi-robot systems, we propose leveraging the abstract behavioral patterns to predict intentions of neighboring agents as critical information for high-speed maneuvers.

By identifying the characteristics of data that birds perceive in flocks, we will develop novel onboard perception strategies specifically designed to achieve fast, distributed UAV swarming in real-world conditions. The bio-insights will be sent to Nature.

Gallery:

In Media:

((Content to be added))

Publications:

((Content to be added))