Page 130 - Human Inspired Dexterity in Robotic Manipulation
P. 130

128   Human Inspired Dexterity in Robotic Manipulation


          grasping force at the instance compression is stopped, there are two possible
          options. One is to give up grasping the object, whereas the other is to
          increase the grasping force (compression) along with the risk of fracture.
          The option of immediately stopping compression upon detection of a linear
          behavior is a little conservative. On the other hand, if compression is allowed
          to continue until the detected linear behavior is over, such a choice is some-
          how dangerous because the end of the linear behavior is close to the fracture
          point. Nevertheless, it must be mentioned that this is not about a grasping
          condition, but a fracture avoidance condition. Frictional and equilibrium
          conditions (balancing of object weight) are additionally required for realiz-
          ing grasping [24]. It must also be remarked that the conditions just described
          apply to materials possessing the same compression behavior as that shown in
          Fig. 7.16, which can generally pertain to gel-like objects.


          7.3.2 Methodology for Detecting Linear Compression
          Behavior for Fracture Avoidance [3]
          7.3.2.1 Compression Test With a Fluid Fingertip
          To avoid fracture, the compression behavior of a fluid fingertip utilized as an
          indenter must be understood. The compression test using a fluid fingertip
          (instead of the indenter in Fig. 7.15) was conducted following the experi-
          mental setup shown in Fig. 7.17. The fluid fingertip is the same as that shown
          in Fig. 7.3 except for the pressure sensor and the control unit. A high pre-
          cision pressure sensor (KEYENCE, AP-12s) was utilized to observe the fluid
          pressure values more accurately. Additionally, a simple control unit was uti-
          lized because its purpose was only to control initial fluid pressures, which
          were set to 1.5, 2.7, 4.0, and 6.0 kPa when there was no contact (i.e., before
          the actual compression test). The experiments were conducted three times




                                              Pump
                           Fluid fingertip
                          x
                                            Load cell
                              f com

                        Target object    Automatic positioning stage

                                           0.0 kPa
          Fig. 7.17 Experimental setup for the compression test using a fluid fingertip on soft tofu.
   125   126   127   128   129   130   131   132   133   134   135