Github: https://github.com/donkarlo/nd_robotic_project
This project aims to empower a flock of aerial swarms to continuously migrate together in search of energy sources (i.e., to recharge their solar batteries—analogous to food) in order to continuously maintain both individual and collective homeostasis.
To achieve this goal, the group must establish an organic relationship with humans and the surrounding environment, enabling bidirectional information exchange through communication, thereby benefiting both parties mutually.
The methodology is grounded in Active Inference, as proposed by Karl Friston. According to this theory, living systems continuously act to minimize surprise by estimating and reducing a latent quantity known as free energy. In this framework, agents maintain their viability by keeping themselves within expected states.
The project is further based on the axiom that any action—whether mental or physical—performed by an individual robot or by the collective inevitably incurs energetic cost and structural wear. Consequently, the ongoing responsibility of each agent, whether biological or artificial, is to minimize surprise (i.e., free energy) while compensating for energy loss and environmental perturbations.