Position and orientation control of an omni-directional mobile rehabilitation robot

Position and orientation control for an omni-directional mobile robot are investigated. The table-placed robot shall be used for arm and shoulder rehabilitation of stroke patients. The position and orientation of the device are determined by means of a modified Kalman filter which encompasses a kinematic model of the robot. By fusing information from incremental encoders at the robot and from an infrared camera at the ceiling, accurate and reliable estimates of robot position and orientation can be obtained. A cascaded position controller is designed for the mobile robot to allow the tracking of arbitrary translational reference movements and the stabilisation of the robot orientation. On the inner control loop, individual angular velocity controllers have been implemented for the three wheels to achieve maximum actuator performance and to facilitate a simplified state-space description of the robot dynamics. For the outer loop, a multi-variable state-space controller is utilised in form of a discrete-time linear quadratic regulator. The sensor fusion scheme and control system are validated in experiments and the obtained results are discussed.