In recent years, robotic applications have been improved for better object manipulation and collaboration with human. With this motivation, the detection of objects has been studied with serial elastic parallel gripper by simple touching in case of no visual data available. A series elastic gripper, capable of detecting geometric properties of objects is designed using only elastic elements and absolute encoders instead of tactile or force/torque sensors. The external force calculation is achieved by employing an estimation algorithm. Different objects are selected for trials for recognition. A Deep Neural Network model is trained by synthetic data extracted from STL file of selected objects . For experimental set-up, the series elastic parallel gripper is mounted on a Staubli RX160 robot arm and objects are placed in pre-determined locations in the workspace. All objects are successfully recognized using the gripper, force estimation and the DNN model. The best DNN model capable of recognizing different objects with the average prediction value ranging from 71% to 98%. Hence the proposed design of gripper and the algorithm achieved the recognition of selected objects without need for additional force/torque or tactile sensors.

Humanoid robot arm design, dynamic and kinematic analyses  2 stages kinesthetic learning algorithm, impedance control and simulation  Suggestions for improvements Figure A. Mechanical design, kinematic and dynamic analysis, electronic design, simulation and control stages for a humanoid robot armPurpose: The purpose of this research is to develop a humanoid robot arm capable of kinesthetic learning and performing tasks which requires contact with environment. Theory and Methods:A humanoid robot arm design and control require maticulus work. Selection of kinematic structure and drivetrain are explained and also verified with dynamic analyses. The inverse kinematic solution is provided. Electronic and software design are realized by keeping the desired control methods in perspective. Simulations are carried out for two-stages kinesthetic learning and position / force / impedance control methods. Results:Although kept relatively simple, the mechanical design has been achived and verified with kinematic and dynamic analyses. The kinesthetic learning is very suitable to teach complicated jobs such as writing on a moving surface. In this project, kinesthetic learning is made possible in two-stages since a single force/torque sensor is available at the end of the robot arm. Harmonic drives' back-drivability becomes an issue during kinesthetic learning. Conclusion:Mechanical design, kinematic and dynamic analyses, kinesthetic learning, impedance control, electronic and software studies were carried out within the scope of the project. The stages from the initial mechanical design of the humanoid robot arm to the control, encountered problems, experiences and suggestions for advanced designs are shared in a comprehensive way in order to be useful for national robot projects in this article.