KSOM based learning for cooperative motion of a redundant rover-manipulator on an uneven terrain

This paper proposes a new learning architecture for the redundant mobile manipulator using Self-Organized Map (SOM), while traversing on an uneven terrain with 3D workspace. The mobile platform considered here is a 10-DOF rover and the manipulator mounted on it has 4 DOR The inverse kinematic map for the 14-DOF redundant mobile manipulator is one-to-many relationship problem. In the proposed approach, a forward kinematic model considering manipulability measure for the manipulator and joint angle constraints is used to generate the data. These data are used to train the network to learn the inverse kinematics relation by a local linear map for each cluster. It produces a unique inverse kinematic solution even for a redundant mobile manipulator and preserves the redundancy in the configuration space. These preserved information is used to generate motions based on tasks, while resolving inverse kinematic problems with a redundant mobile manipulator. The proposed method is able to guide the manipulator end-effector towards the desired target position by minimizing the joint motion. Through simulation experiments, we verified that the proposed SOM based method is found to be effective in generating motion of the mobile manipulator.