Abstract: A model based on multi sensor fusion is proposed for the kinematics planning and control of a redundant manipulator. Pose measuring and inherent characteristics of the sensors are used to fuse information from the sensors in different phases of manipulation, so the planning time is reduced evidently, and the accuracy of pose measuring is increased. Long distance macro planning and near distance micro adjusting are successfully combined to increase grasping stability. By using the geometry characteristics of manipulator, kinematics analysis and singularity avoidance are investigated and analyzed. The method has been applied to our redundant platform, and preferable experimental results have been achieved.