This paper proposes a novel hierarchical motion planning framework for reactive navigation and cooperative manipulation by dual mobile manipulators. When two mobile manipulators grasp an object, they form closed-chain kinematic constraints, which introduce low-dimensional singularities and make the motion planning problem challenging. Moreover, multiple mobile manipulators need to fully consider factors such as redundancy in the manipulators and obstacle information in the environment during the transportation task, as well as the system’s ability to navigate through obstacles. To address the aforementioned issues, we have introduced a hierarchical motion planning framework with integrated reactive motion control. The global planner employs an RRT (Rapidly-exploring Random Tree) sampler to offline plan the motion of the object, effectively calculating various constraints on the path nodes of the object’s movement. The local planner considers reactive planning, allowing for fast and efficient control of the mobile manipulator’s movements within the existing path. We validated the proposed method in simulation, where dual mobile manipulators transport objects towards a target direction in an obstacle-rich environment.