Recently, head-pose estimation has attracted growing attention due to its inherent value in enhancing the performance of head-related applications. Neural networks often require the handling of various representations of the same rotation space. For instance, Euler angles and standard unit quaternion representations can be employed to represent three-dimensional (3D) rotations. However, these representations suffer from the discontinuity problem (DP). Instead of previous representations, we propose a novel end-to-end landmark-free method that employs a continuous quaternion representation (CQR) based on a symmetric matrix ( A ) for 3D rotations when training deep neural networks. This representation is named ContQuat, which satisfies the continuity property and allows the model to overcome the DP and ambiguity issues encountered in commonly used standard unit quaternion and Euler angle representations, enabling efficient full rotation learning. We also introduce three loss functions appropriate for the proposed method to encapsulate the training loss of the neural network during the optimization process. Furthermore, we present a comprehensive experimental comparison of the proposed method against state-of-the-art approaches on publicly available benchmark datasets. The experimental results and error analysis visualizations demonstrate that the proposed method either outperforms or is highly competitive with the current state-of-the-art techniques. The full code is available at: ( ContQuat ).

In this paper, we propose a novel input device for a robotic prosthetic hand based on capacitance change. The proposed device can sense the deformation of the skin due to the activity of the muscle by measuring the capacitance change between the skin and the electrode. Therefore, it can be used as a sensor for estimating a user's intention through medical electrodiagnostic techniques such as electromyogram (EMG)and force myography (FMG). The proposed device can acquire data in a non-invasive way and is advantageous with easier data processing than that for an EMG signal. Moreover, it is resistant to the impedance change of skin because the capacitance is measured in a non-contact manner, unlike the existing methods which work with direct contact with the skin such as EMG. Additionally, unlike FMG, the device is lightly attached to the skin without being strongly fixed with velcro, which offsets problems that occur while re-wearing the device. To demonstrate the feasibility of the proposed idea, three of the newly developed input devices were used to classify four hand movements (fist, scissors, paper, and rest)using a multilayer perceptron (MLP). As a result, the classification success rates for the fist, paper, scissor, and rest motions were obtained as 99.3%, 98.3%, 98.4%, and 99.1%, respectively.

While the central pattern generator (CPG) controller based on biologically inspired neural oscillator networks is known to achieve repeatability and entrainment of rhythmic output motion, no systematic approach to implement the controller has been provided so far. In this study a series of schemes are introduced to tune the control parameters and to set up the network coupling of artificial neural oscillators combined with the virtual force control. Application of the schemes to the proposed controller for a three-link planar robot arm demonstrates that the optimized controller with proper neural oscillator networks can produce minimize the energy consumption with an enhanced-entrained motion with the change of a desired motion and interaction of unknown environments.

Recent progress in wearable exoskeleton robotics results from various needs of an aging society. This would spawn a vulnerable workforce in industry sectors to the near future. Thus, for the application to the industrial field, this work proposes an assistive robotic exoskeleton attached to the calf and thigh of the human leg to totally support a worker's body weight in stand-to-sit with the various seated positions. For this, in this work, a four-bar mechanism based various combinations of passive and active actuators are investigated. Finally the proposed system is composed of one rotary actuator and linear spring based on a four-bar mechanism, considering the high power efficiency for the weight and the mechanical solution to the backdrivability problem of a geared drive. From the extensive simulations, the basic actuation mechanism of the proposed assistive device is confirmed. Then it is verified from experimental results that the robotic exoskeleton can be used to aid who weighs maximum 85 kg in the arbitrary seated position and is actuated with nearly zero current consumption in the walking.

Biologically inspired control approaches have been attracted much attention as alternatives in recent time, for efficiently solving problems in controlling multi-DOF robotic systems, since most human beings or animals exhibit their behaviors in a natural way without explicit computation. Also, they show natural adaptive behaviors irrespective of unexpected external forces or changes of environment. This work is inspired from these novel features. Thus, a self-adapting robotic arm-hand control is proposed exploiting a control scheme based on central pattern generators (CPGs). Instead of a trajectory planning and inverse kinematics problem, this work endeavors to exploit robotic systems coupled with neural oscillators and virtual forces with joint velocity damping. We demonstrate self-adapting motions without the ill-posedness from extensive simulations that enable a robotic arm-hand to make adaptive changes from the given motion to a compliant motion. In addition, it is verified that reaching-to-grasping motion is possible by adopting only transit points sustaining motion repeatability under kinematic redundancy of joints.

This paper proposes a strategy for grasp and manipulation of unknown objects. In order to derive force relations of fingers, the groped shape of soft fingers is introduced. The groped shape is not equal to the real object shape, but is tightly related to the force equilibrium of fingers. By considering contact forces of the groped shape, a simple control parameters can be derived. The manipulation of an object is easily accomplished by embedding the concept of virtual centroid into the grasp control to redistribute internal forces. With these concepts, an object can be easily translated, rotated, and manipulated by relocating fingers. The proposed method is verified with experiments.

This paper presents a development of an anthropomorphic robot hand, `KITECH Hand' that has 4 full-actuated fingers. Most robot hands have small size simultaneously many joints as compared with robot manipulators. Components of actuator, gear, and sensors used for building robots are not small and are expensive, and those make it difficult to build a small sized robot hand. Differently from conventional development of robot hands, KITECH hand adopts a RC servo module that is cheap, easily obtainable, and easy to handle. The RC servo module that have been already used for several small sized humanoid can be new solution of building small sized robot hand with many joints. The feasibility of KITECH hand in object manipulation is shown through various experimental results. It is verified that the modified RC servo module is one of effective solutions in the development of a robot hand.

In recent, control approaches for the human-like behavior in the field of service robotics have been attracting considerable attention, since most humans or animals perform various tasks uncomplicatedly. Hence simple control methods based on gaining a physical insight into human reaching movement in redundancy of DOFs have been proposed. In comparison with the conventional approaches, the proposed method tries to control directly robotic systems in task-space with the control signal composed of linear superposition of three terms 1) joint-damping, 2) virtual spring, and 3) virtual damper in task-space. In particular, our work contains a muscle tension effect of a human under the gravity. This give birth to energy efficient natural motions avoiding problems on repeatability of the motion and ill-posedness problems emerged in most of redundant DOF systems. Thus, this paper exhibits expendability of the position control into the orientation control and compliant behavior. It is verified with a real robotic arm that satisfies human-like movements and motion repeatability under kinematic redundancy of joints.

Redundant manipulator control usually brings about a lot of complexity. This paper proposes a quite simple approach for concurrent set-point regulation of position and orientation of a redundant manipulator using the virtual spring-damper hypothesis, which offers human-like movements without solving the inverse dynamics. For orientation, quaternion representation is applied to give singularity-free orientation control. Augmenting quaternion feedback to the state vector of the system, the stability is analytically proved, and dynamic simulation is conducted for validation of the method. Competition between position and orientation control is also briefly discussed based on the simulation results.

In this paper, we have developed a novel control law to exhibit human-motion characteristics in redundant robot arm-trunk systems for reaching tasks. This newly developed method nullifies the need for the computation of pseudo-inverse of Jacobian while the formulation and optimization of any artificial performance index is not necessary. The time-varying properties of the muscle stiffness and damping as well as the low-pass filter characteristics of human muscles have been modeled by the proposed control law. The newly developed control law uses a time-varying damping shaping matrix and a bijective joint muscle mapping function to describe the human-motion characteristics for reaching motion like quasi-straight line trajectory of the end-effector and symmetric bell shaped velocity profile. The aspect of self-motion and repeatability, which are inherent in human-motion, are also analyzed and successfully modeled using the proposed method. Simulation results show the efficacy of the newly developed algorithm in describing the human-motion characteristics.

Recently, biologically inspired control approaches for robotic systems that involve the use of central pattern generators (CPGs) have been attracting considerable attention owing to the fact that most humans or animals move and walk easily without explicitly controlling their movements. Furthermore, they exhibit natural adaptive motions against unexpected disturbances or environmental changes without considering their kinematic configurations. Inspired by such novel phenomena, this paper endeavors to achieve self-adapting robotic arm motion. For this, biologically inspired CPG based control is proposed. In particular, this approach deals with crucial problems such as motion generation and repeatability of the joints emerged remarkably in most of redundant DOF systems. These problems can be overcome by employing a control based on artificial neural oscillators, virtual force and virtual muscle damping instead of trajectories planning and inverse kinematics. Biologically inspired motions can be attained if the joints of a robotic arm are coupled to neural oscillators and virtual muscles. We experimentally demonstrate self-adaptation motions that that enables a 7-DOF robotic arm to make adaptive changes from the given motion to a compliant motion. In addition, it is verified with real a real robotic arm that human-like movements and motion repeatability are satisfied under kinematic redundancy of joints.

This paper proposes a neural oscillator based control to attain rhythmically dynamic movements of a robot arm. In human or animal, it is known that neural oscillators could produce rhythmic commands efficiently and robustly under the changing task environment. In particular, entrainments of the neural oscillator play a key role to adapt the nervous system to the natural frequency of the interacted environments. Hence, we discuss how a robot arm controls for exhibiting natural adaptive motions as a controller employing the entrainment property. To demonstrate the excellence of entrainment, we implement the proposed control scheme to a real robot arm. Then this work shows the performance of the robot arm coupled to neural oscillators in various tasks that the arm traces a trajectory. Exploiting the neural oscillator and its entrainment property, we experimentally verify an impressive capability of self-adaptation of the neural oscillator that enables the robot arm to make adaptive changes corresponding to an exterior environment.

It is known that biologically inspired neural systems could exhibit natural dynamics efficiently and robustly for motion control, especially for rhythmic motion tasks. In addition, humans or animals exhibit natural adaptive motions without considering their kinematic configurations against unexpected disturbances or environment changes. In this paper, we focus on rhythmic arm motions that can be achieved by using a controller based on neural oscillators and virtual force. In comparison with conventional researches, this work treats neither trajectories planning nor inverse kinematics. Instead of those, a few desired points in task-space and a control method with Jacobian transpose and joint velocity damping are merely adopted. In addition, if the joints of robotic arms are coupled to neural oscillators, they may be capable of achieving biologically inspired motions corresponding to environmental changes. To verify the proposed control scheme, we perform some simulations to trace a desired motion and show the potential features related with self-adaptation that enables a three-link planar arm to make adaptive changes from the given motion to a compliant motion. Specifically, we investigate that human-like movements and motion repeatability are satisfied under kinematic redundancy of joints.

This paper focuses on the Riemannian distance and its application to skilled-motion plannings for the system. The Riemannian distance from one pose to another and vice versa is defined as the minimum curve-length measured by the Riemannian metric based upon the system inertia matrix among all curves connecting the two poses. The minimum-length curve in this meaning is called “geodesic” and reflects a movement of the system affected only by inertia-tensor-originated force (i.e., pure inertia, centrifugal, and Coriolis forces). In order to investigate in detail such a movement along the geodesic, some computer simulations are conducted in the cases of planar motions by a 4-DOF robot arm and biped walkings by a wholebody robot. It is shown through simulation results that movements attaining the Riemannian distance (natural movements in inertial actions) in the two cases tend to be similar to those in human skilled motions when human-scale robot models are chosen. Based upon the Riemannian distance, motion plannings for multi-body systems using physical properties inherent in their own physical structures are discussed.

This paper is concerned with construction of a mathematical model for a class of lumped-parameter dynamics of a pair of robot fingers with soft and deformable tips pinching a rigid object. It is then shown that, in the case of a pair of planer fingers with two and three joints and a 2D rigid object with parallel or non-parallel flat surfaces, there exists a sensory-motor coordinated control signal constructed by using only the knowledge of finger kinematics and measurements of finger joints such that it realizes secure grasping in a dynamic sense. This shows that a pair of robot fingers can grasp a thing securely in a blind manner. The result is further extended to the case of 3D object grasping and manipulation by a pair of soft fingers, one of which can move in 3D space.

A mathematical model expressing motion of a pair of multi-DOF robot fingers with hemi-spherical ends grasping a 3-D rigid object with parallel flat surfaces is derived together with non-holonomic constraints. By referring to the fact that human grasp an object in the form of precision prehension dynamically and stably by opposable forces between the thumb and another finger (index or middle finger), a simple control signal constructed from finger-thumb opposition is proposed and shown to realize stable grasping in a dynamic sense without using object information or external sensing (this is called "blind grasp" in this paper). Stability of grasping with force/torque balance under non-holonomic constraints is analyzed on the basis of the concept called "stability on a manifold". A few of preliminary simulation results is shown to verify the validity of theoretical results

This paper proposes a novel control method for enabling a robotic arm with a pair of two degrees of freedom (DOFs) robotic fingers as an endeffector to execute a variety of superimposed tasks in a coordinated way. Differently from conventional control methods previously proposed for arm-hand systems that adopt a separate design to control each robotic part, such as an arm and a hand, the proposed method designs a unified single control scheme for movements of the whole robotic arm-fingers system. It is assumed that finger-tips are rigid and there are rolling constraints between the finger-tips and a target object, and whole motion of the system is confined to the horizontal plane. The results of numerical computer simulation demonstrate that the proposed method is sufficiently effective in control of such a whole robotic arm-fingers system, and in particular is applicable to superimposed tasks, grasping, regulating, and replacing a target object through composition of the control signal based on the principal of superposition. It is concluded that special attentions in design of control signals should be paid to the wrist that combines the finger parts to the arm

In this paper we propose a control method for pinching an arbitrary shaped object in a horizontal plane by a multi-fingered robot with arbitrary shaped fingertips. It is assumed that the constraints between the fingertips and the object is rolling and smooth, and each finger contacts with the object at one point. The controller requires no object information, no preplanning and no force sensors. It uses only joint angle, joint angular velocity, contact points (tactile feedback) and the fingers' link length. The paper begins with an derivation of dynamics of the finger-object system. We propose a stable grasping controller and briefly show the proof of dynamic stability of the system. The simulation results demonstrate that the controller with tactile feedback can easily realize the stable pinching

In this paper, we study co-activation of digitorum muscles while perform stable pinching and posture regulation tasks of an object by using dual fingers. The fingers have 2 D.O.F. joints and are actuated by nonlinear redundant digitorum muscles to mimic human-like pinching movements. Firstly, we illustrate the kinematics and the dynamics of the overall system, which consider not only the fingers and an object, but also three muscles for each finger to actuate the finger links. Secondly, we consider nonlinear muscle property based on several physiological studies, and propose sensory-motor control rule to the muscles in order to realize stable pinching simultaneously with posture regulation by introducing internal force term induced by co-activation between flexor digitorums and extensor digitorums to modulate the damping factor in joint space. We verify our study by numerical simulations and conclude that this dual fingers system can realize human-like stable pinching and posture regulation

Object manipulation involves two tasks: 1) reaching to grasp and 2) grasping. Motivated by physiological studies related to first task, reaching to grasp, and a simple control method composed of joint-damping and a single virtual spring, we try out for realization of continuous reaching-to-preshaping movements by robotic hand-arm systems with a pair of 2DOFs fingers and a 3DOFs arm. Differently from conventional researches, this paper treats neither trajectories planning nor inverse kinematics. Instead of those, a few transit point and desired point in task-space and a control method with Jacobian transpose and joint velocity damping are merely adopted. Based on results of numerical computer simulation, it is concluded that it is possible to simply realize continuous robotic reaching-to-preshaping motion by adopting the transit points and the control method

Construction of a numerical simulator for 3-D object pinching and manipulation by a pair of robot fingers is described on the basis of mathematical modeling of physical interactions between the object and fingers under rolling constraints. The simulator consists of a set of Lagrange's equations of motion of the fingers and object under holonomic constrains and differential equations of nonholonomic constraints expressing rotational motion of the rigid object. By using this simulator, numerical simulation results of 3-D object pinching and manipulation under the gravity effect based on the control law rendering "blind grasping" without use of any object kinematics or any external sensing is carried out in a case that the fingers have excess joints. Numerical results show the validity of mathematical modeling and the control method

A simple control method for 3-dimensional multi-joint reaching movements under redundancy of Degrees-of-Freedom is proposed, which neither need to introduce any performance index to solve inverse kinematics uniquely nor need to calculate pseudo-inverse of the Jacobian matrix of task coordinates with respect to joint coordinates. The proposed control signal is composed of linear superposition of three terms: 1) angular-velocity feedback for damping shaping; 2) task-space position feedback with a single stiffness parameter; and 3) compensation for gravity force on the basis of estimates for uncertain parameters of the potential energy. Through a variety of computer simulations by using a whole arm model with five DOFs, the importance of synergistic adjustments of damping factors as well as its relation to selection of the stiffness parameter is pointed out. It is shown that if damping factors are chosen synergistically corresponding to the inertia matrix at the initial time and the stiffness parameter then the endpoint converges asymptotically to the target position and reaches it smoothly without incurring any self-motion.

The human thumb is a digit playing crucial roles among all digits used for object manipulation. The special feature of the thumb termed as pad-opposition with four other fingers makes human precision manipulation possible, and during four other fingers dexterously manipulate an object, the thumb functions as a role of stably supporting other's manipulating motions within the pad-opposition. In various characteristics of the thumb, this paper concentrates upon the large radius of pad curvature of the thumb compared with those of four other fingers. In order to show the influence of the large pad curvature, it is shown that reiterated fingering motions can be easily generated by a pair of robotic fingers with different pad curvatures by using a sensory-feedback. It is also shown from computer simulations and experiments that the pad curvature automatically distinguishes roles of the thumb and other fingers. It is finally claimed that the thumb plays a crucial role in enhancement of dexterity and stability in such an implicit manner that it helps the other opposite finger with less pad curvature to make a swift rolling movement that can sophisticatedly control movements of the object.

This paper proposes a manipulation of a circular object in a horizontal plane by a pair of finger robots. This method guarantees the dynamic stability of the system and does not require any object models and object sensing to manipulate the object. It is assumed that there is no friction between the object and the horizontal plane and it is possible to be adequately large friction between the fingertips and the object. We examine the condition of stable grasping of a circular object and propose controllers for stable grasping and for controlling its approximate relative orientation angle without object sensing. The experimental results show the validity of the controller.

This paper proposes a method for controlling an object with parallel surfaces in a horizontal plane by a pair of finger robots. The control method can achieve stable grasping, relative orientation control, and relative position control of the grasped object. The control inputs require neither any object parameters nor any object sensing, such as tactile sensors, force sensors, or visual sensors. The control inputs are also quite simple and do not need to solve either inverse kinematics or inverse dynamics. The stability of the closed-loop system is proved, and simulation and experimental results validate the control method.

In order to enhance dexterity in execution of robot tasks, a redundant number of degrees-of-freedom (DOF) is adopted for design of robotic mechanisms like robot arms and multi-fingered robot hands. However, it is said that surplus DOF of such a robotic mechanism incurs a problem of illposedness of inverse kinematics from task space to joint space. This paper attempts to show that, in the case of a pinching task (grasping in 2D space), by using a pair of robot fingers with hemispherical finger ends, 1) the ill-posedness of inverse kinematics can be resolved in a natural and dynamic way by means of sensory feedback signals from task space to joint space and 2) surplus DOF of fingers enhances dexterity of pinching tasks by remarkably accelerating the speed of convergence toward target object manipulation.

This paper is concerned with a stability theory of motion governed by Lagrange's equation for a pair of multi-degrees of freedom robot fingers with hemispherical finger ends grasping a rigid object under rolling contact constraints. When a pair of two DOF. fingers is used and motion of the overall fingers-object system is confined to a plane, it is shown that the total degree of freedom of the fingers-object system is redundant for realization of stable grasping though there arise four algebraic constraints. To resolve the redundancy problem without introducing extra performance specifications, a concept of stability of motion starting from a higher dimensional manifold to a lower-dimensional manifold expressing a set of states of stable grasp with prescribed contact force is introduced and thereby it is proved in a rigorous way that stable grasp in a dynamic sense is realized by a sensory feedback constructed on the basis of measurement data of finger joint angles and the rotational angle of the object. Further, it is shown that there exists an additional sensory feedback that realizes not only stable grasp but also orientation control of the object concurrently. These results can be extended to other two cases that: 1) motion of the overall system is confined to a vertical plane and therefore it is affected directly by the gravity; and 2) the object has non-parallel but flat surfaces.