Intelligence Through Interaction: Leveraging Physical Interactions In Simple Robots To Produce Complex Behaviors

Abstract

This thesis investigates how intelligent robot behavior can emerge from physical interactions rather than sensing, computation, and actuation in the traditional sense. Two robotic systems are presented to explore this concept in different domains. The first is a swarm of simple rolling robots whose collective morphology is shaped by distributed control laws and magnetic interactions, enabling decentralized construction-like behaviors such as bridge formation. The second is a soft underwater robot inspired by anguilliform swimming, which achieves efficient locomotion through a single actuator that leverages fluid–structure interactions in a compliant silicone tail. Useful behavior arises in both systems from the physical design and the dynamics of environmental interaction, rather than from algorithmic or computational complexity. These results demonstrate that physical intelligence can serve as a powerful design principle for building adaptive, robust, and minimal robotic systems.