Programmable matter (PM) is a form of matter capable of dynamically altering its physical properties, such as shape or density, through programmable means. From a robotics perspective, PM can be realized as a distributed system consisting of numerous small computational entities working collaboratively to achieve specific objectives. Although autonomous mobile robot systems serve as an important example and have been researched for more than two decades, these robots often fail to perform even basic tasks, revealing a considerable gap in PM implementation. In this paper, we introduce a novel computational paradigm, termed the Pairing Robot model (Pairbot model), which is built on an autonomous mobile robot system. In this model, each robot forms a pair with another, enabling them to recognize each other and adapt their positions to achieve designated goals. This fundamental principle of pairing substantially enhances inter-robot connectivity compared to conventional LCM-type model, even under asynchronous scheduler conditions. This shift has considerable implications for computational capabilities, specifically in problem solvability. We explore two specific challenges -- the perpetual marching problem and the 7-pairbots-gathering problem -- to demonstrate the computational power of Pairbot model. This model provides new avenues and insights to address inherent issues in autonomous mobile robots.