Resilient BotController Strategies for Modular Cubic Robot Clusters

This technical note presents a resilience-first approach to the control of modular robots and programmable matter. Because such systems cannot assume ideal behavior from every bot, simulators must support targeted and reproducible fault injection so that the BotController can detect, classify, and respond to failures in a principled way. We introduce an initial taxonomy of 18 fault types covering World Model Drift, Obstruction and Movement Failure, Communication Faults, and Access and Identity Faults. For each class, we outline how the corresponding faults can be injected in the simulator and how an LLM-assisted BotController can identify and handle them using available diagnostic tools. Rather than aiming at exhaustive evaluation, this note provides a structured starting point for fault-aware simulation, robust control, and recovery workflows in SP-CellBots and related modular robotic systems.