• The Journal of Korea Robotics Society
• /
• v.5 no.2
• /
• pp.102-109
• /
• 2010

In this study, an anthropomorphic robot Hand, called "SKKU Hand III" is presented. The hand has thirteen DOF(Degree-Of-Freedom) and is designed based on the skeletal structure of the human hand. Each finger module(except thumb module) has three DOF and four joints with a saddle joint mechanism which has two DOF at the base joint. Two distal joints of the finger module are mechanically coupled by a timing belt and pulleys. The thumb module is composed of a finger module and an additional actuator, which makes it possible to realize the opposition between the thumb and the other fingers. In addition, the palm DOF of the human hand is mimicked with a spatial link mechanism between the index finger and the thumb. Thus, it can grasp objects more stably and more strongly. For the modularization of the robotic hand all the driving circuits are embedded in the hand, and only the communication lines supporting CAN protocol with DC power cable are given as an interface. Therefore, it is possible to apply it to any robot system the interface. To validate the feasibility of the SKKU Hand III, a series of the representative grasp experiments such as power, precision, intermediate grasp etc. are carried out with the object around us and its operation is demonstrated.

• ETRI Journal
• /
• v.38 no.6
• /
• pp.1218-1228
• /
• 2016

This paper proposes a robot hand for a violin-playing robot and introduces a newly developed robot finger. The proposed robot hand acts as the left hand of the violin-playing robot system. The violin fingering plays an important role in determining the tone or sound when the violin is being played. Among the diverse types of violin fingering playing, it is not possible to produce vibrato with simple position control. Therefore, we newly designed a three-axis load cell for force control, which is mounted at the end of the robot finger. Noise is calculated through an analysis of the resistance difference across the strain gauge attached to the proposed three-axis load cell. In order to ensure the stability of the three-axis load cell by analyzing the stress distribution, the strain generated in the load cell is also verified through a finite element analysis. A sound rating quality system previously developed by the authors is used to compare and analyze the sound quality of the fourth-octave C-note played by a human violinist and the proposed robot finger.

• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.5
• /
• pp.179-186
• /
• 2021

Manipulation of complex objects with an anthropomorphic robot hand like a human hand is a challenge in the human-centric environment. In order to train the anthropomorphic robot hand which has a high degree of freedom (DoF), human demonstration augmented deep reinforcement learning policy optimization methods have been proposed. In this work, we first demonstrate augmentation of human demonstration in deep reinforcement learning (DRL) is effective for object manipulation by comparing the performance of the augmentation-free Natural Policy Gradient (NPG) and Demonstration Augmented NPG (DA-NPG). Then three DRL policy optimization methods, namely NPG, Trust Region Policy Optimization (TRPO), and Proximal Policy Optimization (PPO), have been evaluated with DA (i.e., DA-NPG, DA-TRPO, and DA-PPO) and without DA by manipulating six objects such as apple, banana, bottle, light bulb, camera, and hammer. The results show that DA-NPG achieved the average success rate of 99.33% whereas NPG only achieved 60%. In addition, DA-NPG succeeded grasping all six objects while DA-TRPO and DA-PPO failed to grasp some objects and showed unstable performances.

• Journal of Computing Science and Engineering
• /
• v.7 no.2
• /
• pp.122-131
• /
• 2013

The MPI CyberMotion Simulator provides a unique motion platform, as it features an anthropomorphic robot with a large workspace, combined with an actuated cabin and a linear track for lateral movement. This paper introduces the simulator as a tool for studying human perception, and compares its characteristics to conventional Stewart platforms. Furthermore, an experimental evaluation is presented in which multimodal human control behavior is studied by identifying the visual and vestibular responses of participants in a roll-lateral helicopter hover task. The results show that the simulator motion allows participants to increase tracking performance by changing their control strategy, shifting from reliance on visual error perception to reliance on simulator motion cues. The MPI CyberMotion Simulator has proven to be a state-of-the-art motion simulator for psychophysical research to study humans with various experimental paradigms, ranging from passive perception experiments to active control tasks, such as driving a car or flying a helicopter.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2000.08a
• /
• pp.113-116
• /
• 2000

This paper analyzes the singularity of an anthropomorphic robot associated with joint and operational stiffness generation from muscle stiffness. The singularity analysis is made simply based on the signs of the actual and the desired coupling joint stiffness. First, the relationships of the muscle stiffness and the actual joint stiffness, and the operational stiffness and the desired joint stiffness are examined. Second, according to the sign restriction on the actual coupling joint stiffness, the operational space is divided into the semi-singular(SS), the regular(R), and the semi-regular(SR) regions. Third, from the sign comparison of tile actual and the desired coupling joint stiffness, the sufficient condition for the semi-singularity in operational stiffness generation is derived. The limitation on the allowable operational stiffness when a task point belongs to SS, R, and SR regions is also discussed. Simulation results are given.

• International Journal of Control, Automation, and Systems
• /
• v.4 no.2
• /
• pp.227-235
• /
• 2006

For a sophisticated humanoid that explores and learns its environment and interacts with humans, anthropomorphic physical behavior is much desired. The human vision system orients each eye with three-degree-of-freedom (3-DOF) in the directions of horizontal, vertical and torsional axes. Thus, in order to accurately replicate human vision system, it is imperative to have a simulator with 3-DOF end-effector. We present a 3-DOF anthropomorphic oculomotor system that reproduces realistic human eye movements for human-sized humanoid applications. The parallel link architecture of the oculomotor system is sized and designed to match the performance capabilities of the human vision. In this paper, a biologically-inspired mechanical design and the structural kinematics of the prototype are described in detail. The motility of the prototype in each axis of rotation was replicated through computer simulation, while performance tests comparable to human eye movements were recorded.

• Science of Emotion and Sensibility
• /
• v.24 no.4
• /
• pp.3-16
• /
• 2021

The attribution of human traits, emotions, and intentions to nonhuman entities such as robots is known as anthropomorphism. Two studies were conducted to check whether human-robot interaction is affected by anthropomorphic framing of robots. In Study 1, participants were presented with pictures of robots that varied in human similarity in appearance. According to the results, uncanny feelings toward a robot increased with the higher levels of human similarity. Furthermore, as the level of mind attribution increased, participants tended to attribute more humanlike abilities to nonhuman agents. In Study 2, a robot was described as either a machine-like robot or a humanlike robot in a priming story; then, it was examined whether significant differences exist in mind attribution and moral care. The participants tended to perceive robots as more humanlike in the mind attribution when anthropomorphism was used in a robot's behavior, according to the findings. Furthermore, in the condition of increased anthropomorphism, a higher level of moral care could be observed compared with that in the other condition. This means that humanlike appearances may increase uncanny feelings, whereas anthropomorphic attribution may facilitate social interactions between humans and robots. Limitations as well as the implications for future research are discussed.

• Proceedings of the Korean Society of Machine Tool Engineers Conference
• /
• 1999.10a
• /
• pp.131-136
• /
• 1999

This paper proposes the robot inverse calibration method using a neural networks. A highorder networks called Pi-Sigma networks has been used. The Pi-Sigma networks uses linear summing units in the hidden layer and product unit in output layer. The inverse calibration model which compensates the difference of joint variables only between measuring value and analytic value about the desired pose(position, orientation) of a robot is proposed. The compensated values are determined by using the weights obtained from the learning process of the neural networks previously. To prove the reasonableness, the SCARA type direct drive robot(4-DOF) and anthropomorphic robot(6-DOF) are simulated. It shows that the proposed calibration method can reduce the errors of the joint variables from $\pm$2$^{\circ}$to $\pm$ 0.1$^{\circ}$.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.19 no.1
• /
• pp.181-192
• /
• 1995

In this study, we introduce new types of planar 2 degree-of-freedom robot modules resembling the musculoskeletal structure of the human arm with actuation redundancy. First, for the given actuator sizes the performance analysis for the manipulator with redundant actuation and without redundant actuation is performed with respect to maximum load handling capacity, maximum hand velocity, and maximum hand acceleration. Secondly an algorithm which decides optimal actuator sizes for the given operational performances is introduced, and the optimal actuator sizes for a robot module with four redundant actuation are obtained. The algorithms employed in this paper will be useful to analyze the robot performances and to determine the actuator sizes for general robot manipulators.

• Proceedings of the Korean Society of Machine Tool Engineers Conference
• /
• 1999.05a
• /
• pp.199-204
• /
• 1999

The robot inverse calibration method using a neural networks is proposed in this paper. A high-order networks has been used in this study. The Pi-Sigma networks uses linear summing units in the hidden layer and product unit in output layer. The inverse calibration model which compensates the difference of joint variables only between measuring value and analytic value about the desired pose(position orientation) of a robot is proposed. The compensated values are determined by using the weights obtained from the learning process of the neural networks previously. To prove the reasonableness, the selected compliance automatic robot arm type direct drive robot and anthropomorphic robot are simulated. It shows that the proposed calibration method can reduce the errors of the joint variables from ${\pm}$0.15$^{\circ}$to ${\pm}$0.12$^{\circ}$.