A systematic method for an autonomous decentralized control system is still lacking, despite its appealing concept. In order to alleviate this, we focused on the amoeboid locomotion of the true slime mold, and extracted a design scheme for the decentralized control mechanism that leads to adaptive behavior for the entire system, based on the so-called discrepancy function. In this paper, we intensively investigate the universality of this design scheme by applying it to a different type of locomotion based on a 'synthetic approach'. As a first step, we implement this design scheme to the control of a real physical two-dimensional serpentine robot that exhibits slithering locomotion. The experimental results show that the robot exhibits adaptive behavior and responds to the environmental changes; it is also robust against malfunctions of the body segments due to the local sensory feedback control that is based on the discrepancy function. We expect the results to shed new light on the methodology of autonomous decentralized control systems.