Many species such as eels, lampreys and leeches generate undulatory swimming locomotion adaptively. It is said that this coordinated locomotive patterns are produced by central pattern generators (CPGs) which generate rhythmic activities without any rhythmic inputs. Additionally, there are some local sensors underlying in their bodies (e.g. lampreys: stretch receptors, larval zebra-fish: lateral organs). We assumed that such several sensors likely cooperate and influence their adaptive locomotion with CPGs. However, there is still very little understanding how CPGs and multimodal local sensors interact for adaptive locomotive patterns. In this study, we aim to design a minimal CPG model for a swimming robot with multimodal local sensory feedback which can produce an adaptive undulatory swimming locomotion. Finally, we validated it under different conditions via 2D simulation.