This paper deals with motion planning for a multifunctional underwater robot which can accomplish various missions, e.g., swimming, walking and grasping objects. Authors have developed a unified motion planning method which can generate motion planning for a variety of task by a single algorithm. In this method, motion planning problems are modeled as finite horizon Markov decision processes. The optimum motion planning is obtained by Dynamic Programming, however Dynamic Programming is sometimes thought to be of limited applicability because of the curse of dimensionality. To avoid the curse of dimensionality, authors applied a random network as a state transition network. The explosion of the number of states can be suppressed by using the random network. The effectiveness of the proposed method is demonstrated through numerical simulations of two types of tasks for multifunctional robots. One is a reaching task, the other is a generating thrust force task.