Core Operating System for Autonomous Robots: Deterministic Action without Optimization via Structural Elimination

This study introduces a core operating system for autonomous robots in which actions are determined deterministically through structural elimination, without reliance on optimization. The proposed framework initiates autonomous action through impulse-driven activation and structured elimination over a finite modular action library. Unlike conventional approaches that depend on optimization, learning, or explicit goal specification, the system operates through the accumulation of discrete event impulses that activate constraint structures, progressively eliminating incompatible actions until a consistent outcome is revealed. Actions are not selected through search or ranking; instead, incompatible candidates are eliminated until a single admissible element remains. A threshold-based activation mechanism governs the transition from passive observation to active resolution, ensuring that decisions are triggered only when sufficient structural information is available. The framework introduces a hierarchical modular library architecture that supports both routine behavior and event-driven extension. New action patterns are incorporated through structured recombination when existing configurations fail to resolve emerging conditions, enabling continuous yet controlled system evolution without parameter learning. The same impulse–accumulation–threshold–elimination mechanism operates consistently across scales, from local adjustments to global behavioral reconfiguration. The framework further extends to multi-agent settings through selective library exchange, allowing collaborative systems to evolve without centralized control. The results establish a complementary paradigm for autonomous systems in which decision-making emerges from feasibility and structural consistency rather than optimization, offering a lightweight, interpretable, and deterministic alternative for structured environments.