To deal with the problem of obstacle avoidance for redundant dualarm robot, an obstacle avoidance algorithm based on selfmotion was presented. In order to fully describe a cooperative task for a dualarm robot system, this allows a straightforward description of general coordinated motion in terms of meaningful absolute and relative variables. The proposed Jacobians were expressed in terms of the Jacobians of the single manipulators. Based on this, the kinematics inverse solution of redundant dualarm robot used to avoid obstacles was obtained. Using this approach, the robot can accomplish the coordinated motion tasks while avoiding the obstacles. Finally, the validity of the algorithm was manifested by simulation of a redundant dualarm robot. Results show that the closest distance between the manipulator and the obstacle was more than 0.015m, while the endeffector of the robot can accomplish the coordinated motion tasks accurately and the path was smooth. As a conclusion, the redundant robot can not only realize the obstacle avoidance, but also accomplish the coordinated motion using the proposed approach.