A mask area based grasping force control strategy for force sensor-less robot

The grasping force of service robots should be controlled within a proper range when they operate objects with unknown physical property (OUPP) whose physical property may be deformable. And that is challenging for service robots equipped with force sensor-less gripper. Therefore, this paper introduced a mask area based grasping force control strategy (MAFC) to realize the control of the grasping force in visual mode. Firstly, a semantic segmentation model was applied to monitor the deformation of the object. And this deformation was treated as a criterion to conduct the evaluation of the grasping state. Then the gripper can be adjusted based on the grasping state. Besides, a cup grasping experiment was conducted with the MAFC strategy. The experimental results showed that the success rate of grasping can be 90%. Meanwhile, the proportion of deformation was controlled within 2%. Moreover, with the MAFC strategy, the contrast experiments indicated that the grasping success rate can be increased by 40% compared with the” Pick and Place” module of MoveIt. All in all, the MAFC strategy can improve the grasping performance of service robots with force sensor-less gripper.