Locomotion and self-reconfiguration autonomy for spherical freeform modular robots

Abstract

Modular robotic systems are multi-robot systems comprising numerous repeated modules and can transform into different configurations. Matching system configurations to a library enables efficient automation of modular robotic systems that have high degrees of freedom and strict motion constraints. Many previous approaches have automated cube-oriented modular robots by mapping the predefined sequence of gaits in the library to the module controllers. However, they can hardly drive robust three-dimensional self-reconfigurations without external sensors due to limited gait control accuracy and docking misalignment tolerance. Freeform modular robots are a type of modular robot with no fixed-point connectors, typically featuring continuous spherical joint connections between modules. They exhibit higher docking misalignment tolerance and better environmental adaptability. However, existing library-driven systems are inapplicable to freeform robots due to their redundant degrees of freedom and incompatible self-reconfiguration approaches. This article first proposes an autonomy framework for the locomotion and self-reconfiguration of spherical freeform modular robots. We model module connections as either spherical joints or parallel robots, employing a unified approach for skeletal kinematics. The system achieves the target configuration through iterative inverse kinematics and command translation to module controllers. A library with interfaces for configuration design is proposed, defining behaviors and feasible kinematic transitions between configurations. The executable behavior can be efficiently retrieved from the library by combining the proposed configuration matching and mapping algorithm. The system is validated on the FreeSN system with up to 18 modules containing 48 joint motors, providing a foundation for high-level planning and control research in freeform modular robots.