International Journal of Humanoid Robotics

Starting from the operational space and task prioritization framework, presented in [L. Sentis and O. Khatib, Task-oriented control of humanoid robots through prioritization, in Proc. IEEE-Robotics and Autonomous Systems/RSJ International Conf. Humanoid Robots, Santa Monica, CA, USA, November 2004.], this paper proposes an extension and improvement of this approach, to make it applicable to nonholonomic tasks and systems. For the tasks where inequality type conditions have to be fulfilled, such solutions are obtained to ensure as small as possible movements at the joints, while keeping the actuators' driving torques between saturation limits. Having in mind that a prerequisite for realization of any task by biped robot is the maintenance of its upright position, this issue is also in the focus of our study. Instead of keeping the zero-moment point (ZMP) at an exact position, dynamic balance was ensured by allowing the ZMP to be anywhere within the support area. Since the condition for ZMP position is relaxed a smaller number of joints are engaged in the task realization, which enables more tasks to be handled simultaneously. Simulations were performed, and the results proved the validity of the proposed approach. When disturbance was applied compensation behavior emerged.

In this paper, a new approach to SLAM is proposed that is based on particle filter and soft computing techniques. In this approach, the robot pose is estimated based on unscented marginal particle filter (UMPF) and the static map is considered as parameters that are updated using soft computing. Significant improvement in the proposed method is observed in terms of accuracy of estimation and consistency compared to conventional methods. A number of simulations and experiments are presented to evaluate the algorithm's performance compared to conventional approaches.

In this paper, we present a vision-based localization system using mobile augmented reality (MAR) and mobile audio augmented reality (MAAR) techniques, applicable to both humans and humanoid robots navigation in indoor environments. In the first stage, we propose a system that recognizes the location of a user from the image sequence of an indoor environment using its onboard camera. The location information is added to the user's view in the form of 3D objects and audio sounds with location information and navigation instruction content via augmented reality (AR). The location is recognized by using the prior knowledge about the layout of the environment and the location of the AR markers. The image sequence can be obtained using a smart phone's camera and the marker detection, 3D object placement and audio augmentation will be performed by the phone's operating processor and graphical/audio modules. Using this system will majorly reduce the hardware complexity of such navigation systems, as it replaces a system consisting of a mobile PC, wireless camera, head-mounted displays (HMD) and a remote PC with a smart phone with camera. In the second stage, the same algorithm is employed as a novel vision-based autonomous humanoid robot localization and navigation approach. The proposed technique is implemented on a humanoid robot NAO and improves the robot's navigation and localization performance previously done using an extended Kalman filter (EKF) by presenting location-based information to the robot through different AR markers placed in the robot environment.

Even though many humanoid robots have been developed and they have locomotion ability, their balancing ability is not sufficient. In the future, humanoid robots will work and act within the human environment. At that time, the humanoid robot will be exposed to various disturbances. This paper proposes a balancing strategy for hopping humanoid robots against various magnitude of disturbance. The proposed balancing strategy for a hopping humanoid robot consists of two controllers, the posture balance controller and the landing position controller. The posture balance controller is used for small disturbances, and its role is to maintain stability by controlling the ankle torque of the robot. On the other hand, if disturbance is large, the landing position controller, which changes the landing position of the swing foot, works with the posture balance controller simultaneously. In this way, the landing position controller reduces large disturbances, and the posture balance controller controls the remaining disturbances. The landing position controller is derived by the principle of energy conservation. An experiment conducted with a real humanoid robot, HUBO2, verifies the proposed method. HUBO2 made a stable and continuous hopping action with the proposed balancing strategy overcoming various disturbances placed in the way of the robot.

Biped humanoid robots have been developed to successfully perform human-like locomotion. Based on the use of well-developed locomotion control systems, humanoid robots are further expected to achieve high-level intelligence, such as vision-based obstacle avoidance navigation. To provide standard obstacle avoidance navigation problems for autonomous humanoid robot researches, the HuroCup League of Federation of International Robot-Soccer Association (FIRA) and the RoboCup Humanoid League defined the conditions and rules in competitions to evaluate the performance. In this paper, the vision-based obstacle avoidance navigation approaches for humanoid robots were proposed in terms of combining the techniques of visual localization, obstacle map construction and artificial potential field (APF)-based reactive navigations. Moreover, a small-size humanoid robot (HuroEvolution^JR) and an adult-size humanoid robot (HuroEvolution^AD) were used to evaluate the performance of the proposed obstacle avoidance navigation approach. The navigation performance was evaluated with the distance of ground truth trajectory collected from a motion capture system. Finally, the experiment results demonstrated the effectiveness of using vision-based localization and obstacle map construction approaches. Moreover, the APF-based navigation approach was capable of achieving smaller trajectory distance when compared to conventional just-avoiding-nearest-obstacle-rule approach.

Push recovery is an important capability for a biped to safely maneuver in a real dynamic environment. In this paper, a generalized push recovery scheme to handle pushes from any direction that may occur at any walking phase is developed. Using the concept of walking phase modification, a series of systematic push recovery scheme that takes into account the severity of the push is presented. The result is that a bipedal robot could adapt to pushes according to the magnitude of disturbance and determine the best course of action. A number of push recovery experiments with different walking phases and push directions have been carried out using a 12-DOF humanoid robot model in dynamic simulations. The versatility and potential of the overall scheme is also demonstrated with the bipedal robot balancing on an accelerating cart.

Kinematic efficiency of robot limbs can be analyzed by means of the manipulability concept defined by Yoshikawa. However, when it is applied to redundant serial chains, that are the most common case in humanoid robotics, Yoshikawa's manipulability measure is not able to take into account the joint velocity range of each robot joint. In order to overcome this difficulty we propose to substitute in the place of the manipulability ellipsoid a manipulability zonotope resulting from the robot Jacobian transform of the joint velocities parallelotope. The proposed manipulability zonotope notion is similar to the well-known manipulability polytope notion, but its associated volume measure can take benefit of the so-called zonotope theory. By means of a famous and powerful theorem about the cubical dissection of any zonotope, an original formula of manipulability measure is proposed from which is also derived a mean manipulability estimate associated to a given joint space working. The proposed tools are applied to a kinematic analysis of a 4R-regional structure model of the upper limb.

Children with autism spectrum disorders have difficulties in identifying situation-based emotions, which is a fundamental ability for mind reading. Social robots received increased attention as assisting tools for improving the social and emotional skills of children with autism. This study investigates whether the social robot Probo can help children with autism spectrum disorders to enhance their performance in identifying situation-based emotions. Three participants (age between 5 and 6) diagnozed with autism spectrum disorders were included in a single case AB experimental design, with intersubjects replications. The results show that children's performance improved with moderate to large effect sizes in identifying both sadness and happiness. Based on these results, we intend to perform more extensive investigations regarding the effectiveness of robot assisted therapy in improving social-emotional abilities for children with autism spectrum disorders.

This paper presents a method based on virtual finger (VF) set to analyze the grasp function of bionic hand. In order to exactly represent different function units of human hand or bionic hand a more detailed classification of VF is proposed to divide VF into three types: VF_FE, VF_AA and VF_Palm, of which a VF set describing hand grasp can be composed. The VF set cannot only describe the degree of freedom (DOF) distribution of human hand or bionic hand but also indicate which fingers take part in grasp and what motion types the fingers are. The VF sets of eight classic bionic hands and five basic human hand grasp postures are provided. The inclusion relationship between the VF set of a bionic hand and those of the basic human hand grasp postures is used to analyze the grasp function of a bionic hand. As examples the grasp functions of several classic bionic hands, such as i-LIMB Hand, HIT/DLR Hand, etc. are analyzed.

The force distribution problem (FDP) in robotics requires the determination of multiple contact forces to match a desired net contact wrench. For the double support case encountered in humanoids, this problem is underspecified, and provides the opportunity to optimize desired foot centers of pressure (CoPs) and forces. In different contexts, we may seek CoPs and contact forces that optimize actuator effort or decrease the tendency for foot roll. In this work, we present two formulations of the FDP for humanoids in double support, and propose objective functions within a general framework to address the variety of competing requirements for the realization of balance. As a key feature, the framework is capable to optimize contact forces for motions on uneven terrain. Solutions for the formulations developed are obtained with a commercial nonlinear optimization package and through analytical approaches on a simplified problem. Results are shown for a highly dynamic whole-body humanoid reaching motion performed on even terrain and on a ramp. A convex formulation of the FDP provides real-time solutions with computation times of a few milliseconds. While the convex formulation does not include CoPs explicitly as optimization variables, a novel objective function is developed which penalizes foot CoP solutions that approach the foot boundaries.

This paper presents an anthropomorphic prosthetic hand, which has a thumb and four fingers, all the fingers of which are driven by servomotors built into the fingers and the palm. A novel flexible three-axis tactile sensor array have been investigated for their suitability to measure the grip forces exerted upon an object held by a prosthetic hand. The tactile sensor array are placed at the thumb of prosthetic hand with the capability of measuring both normal and shear force distribution using quantum tunneling composite (QTC) as a base material. A joystick-like mesa was attached to a sensor base to transfer external force, and there are four fan-shaped electrodes in a cell to decompose the contact force into normal and shear components. The sensor has been realized in a 2 × 6 array of unit sensors, and each unit sensor responds to normal and shear stresses in all three axes, respectively. The zero potential method is used to avoid crosstalk effect by setting all drive lines (row electrodes), which do not involve the measuring point at zero voltage and also by setting all output lines (column electrodes) at zero voltage. Measurement of a single sensor shows that the full-scale range of detectable force are about 10, 10 and 22 N for the x-, y- and z-directions, respectively. The sensitivities of a cell measured with a current setup are 0.47, 0.45 and 0.16 mV/mN for the x-, y- and z-directions, respectively. The sensor showed a high repeatability, low hysteresis, and min tactile cross-talk. The proposed flexible three-axial tactile sensor array can be applied in a curved or compliant surface that requires slip detection and flexibility, such as a prosthetic hand.