This paper deals with the pneumatic gripper only using On-off valve grasping manipulation by using Neural network. A position control for gripper is proposed with consideration of the non-linearities of the pneumatic system. This proposed controller generates a "Switching time" to provide the "Ideal motion" of the robot hand. The grasp force control system shows its intelligence that absorbs the human's operating experiences and thinking methods. The performance of fabricated pneumatic gripper is investigated experimentally for the reduction of impact force and grasping time, the ability of generating grasp force to provide stable grasp state. Experimental results show that the level of enhanced performance of the pneumatic gripper.