Journal of Pioneering Artificial Intelligence Research

Abstract

This study introduces a core operating system for autonomous robots in which actions are determined de terministically through structural elimination, without reliance on optimization. The proposed framework initiates autonomous action through impulse-driven activation and structured elimination over a finite mod ular 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 admis sible element remains. A threshold-based activation mechanism governs the transition from passive observa tion 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 frame work further extends to multi-agent settings through selective library exchange, allowing collaborative sys tems 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 optimiza tion, offering a lightweight, interpretable, and deterministic alternative for structured environments.