Simulation and modular controller integration for self-reconfigurable robots with docking/undocking behaviors

Abstract

Modular self-reconfigurable robots (MSRRs) offer structural adaptability for diverse tasks and environments. However, simulating such robots presents challenges in handling dynamic topology and closed-loop connections. PhysX, a modern high-performance physics engine widely adopted in robotics simulators including Isaac Sim, provides efficient and stable simulation through its articulation system for tree-structured robots. Yet, this articulation mechanism inherently prohibits closed-loop topologies, which are common in MSRRs. To address this, we propose a hybrid simulation and control framework for MSRRs. First, we propose the Soft-Constraint Joint (SCJ), a constraint mechanism implemented via articulation-excluded joints that enables loop formation and dynamic module connection without violating articulation assumptions. Second, we present a modular control architecture where each controller module encapsulates motion logic and can be automatically activated based on physical configuration. The proposed framework supports stable and scalable simulation of modular robots with seamless integration of perception, reconfiguration, and control. Experiments validate the effectiveness of the framework in handling multiple robot morphologies and in achieving reliable sim-to-real transfer.