In this study, we propose a model predictive obstacle avoidance control for leg/wheel mobile robots with optimized articulated leg configuration. Using the proposed method, the robot can avoid obstacles and keep stability in narrow and complex spaces like indoor environments while the robot allocates wheels adapted to the surrounding environment. We optimize the angular velocities of joints and wheel driving speeds using model predictive control. Although our past related researches optimize the joint angles and wheel speeds separately, the proposed method optimizes these simultaneously. Thus, enhanced optimality of the entire robot system is expected to be achieved. We conduct numerical simulations of an actual mobile robot to verify efficacy and feasibility of the proposed method, which demonstrate that the proposed method is effective to avoid obstacles.