Design, realization and control of biomedical-soft robot hand

In recent years, soft robots, a new field of robotics, have appeared in many areas. The most important issues of soft robots are the designs and controls of soft robots. To overcome the difficulties mentioned in this study, soft-finger models with 13 different pneumatic systems have been designed in the Solidworks drawing program. Static analyzes of the finger designs have been carried out in the ANSYS package program to assess the finite element method (FEM). It was examined how the amount of strain, stress, and deformation occurring in the fingers was affected when the amount of pressurized air was sent to the fingers, and the design parameters were changed. The obtained static analysis results are trained on an artificial neural network (ANN) and a mathematical model has been created. Thus, the ANN model has obtained estimated results of strain, stress, and deformations. Thus, estimated results of strain, stress and deformations were obtained with the ANN model. When the actual results obtained from ANSYS and the estimated results of ANN are compared, MSE, RMSE, MEA, R, NSE and R2 error values are calculated as 0.07%, 1.87%, 2.14%, 99.51%, 97.67% and 99.60%, respectively. In line with these results, the best model has been determined as the 2nd model. In addition, ANN and extreme learning machine (ELM) models have been created to assess the estimation of the required pressure value by using the deformation and time parameters obtained as a result of static analyses. Error-values of these models have been compared. As a result of the error values in the ANN model and the ELM model for pressure value estimation, the best estimate is obtained from the 8th design. Taking into account the data obtained as a result of the analysis, a soft robotic hand with 5 fingers has been designed experimentally. This hand is made of Dragon Skin-20 material, which is a soft material, with the thumb having 6 channels and the other fingers having 11 channels. Grasping objects of variable thickness has been achieved by controlling the air pressure in the fingers of the soft robot hand.