Kinodynamic motion planning for legged mobile manipulator with large object's dynamics

AbstractThe extension of traditional manipulators' workspace through mobile platforms has broadened the spectrum of potential robotic applications. Legged mobile manipulators, in particular, have garnered increasing attention due to their capability to navigate diverse environments and tackle challenging terrains more effectively than the wheeled counterparts. This study investigates the intricacies of motion planning for legged mobile manipulators in the context of manipulating large objects. Legged manipulators can significantly enhance their ability to manipulate large objects through coordinated movements of their legs and bodies. However, the complexity of this task far surpasses that of ordinary mobile manipulation challenges. To address this complex problem, it is decomposed into two sub-problems, focusing on the dynamics of both the object and the robot. The direct collocation method is used to transform the original problem into nonlinear programming, and suitable initial values are obtained using whole-body inverse kinematics. Finally, we demonstrate robot's ability to perform tasks such as opening a heavy door and putting a tumbled chair upright, thereby illustrating the efficacy and practicality of our approach.