Robots have been shown to assist in education and development of social skills in children. However, there has been no study yet that benchmarks the effectiveness of robots with respect to traditional playtools found inside a classroom, such as pretend play items and blocks. This paper presents the design, development and testing of robotic puppets, which would be used to support teaching in kindergarten education. Different types of robotic puppet design were considered before settling on a glove-type puppet. To benchmark the robot performance, a total of 52 five year-old children were observed, from which quantitative and qualitative data were collected. The result of the study indicates that when playing with the robotic puppets, the performance of the children with respect to thinking and learning, creativity and imagination, and social interaction and independence, is comparable to other traditional playtools.

A manipulator with variable stiffness allows the manipulator to adjust its stiffness to fulfill different task requirements. In this paper, a cable-driven manipulator with the ability to significantly regulate its stiffness through tension manipulation is introduced. Variable stiffness is achieved by attaching a novel variable stiffness device along each driving cable, in which the stiffness of the device is a function the cable tension. As cable-driven manipulator has actuation redundancy, the tension distribution can be manipulated even at a stationary pose. Such property allows the cable-driven manipulator to adjust the stiffness of each variable stiffness device, thereby changing the stiffness of the manipulator. The design and analysis of the variable stiffness device is presented. The variable stiffness device uses commercial torsion springs, and has a compact and light-weight design. Experimental and simulation results verified that cable-driven manipulator with such variable stiffness devices is able to achieve significant stiffness regulation.

The paper describes a Yoga training system that is built based on motion replication technique (MoRep), including hardware, virtual scenario and feedback design. The motion replication technique proposed here can determine the similarity between Yoga master and student's postures and then provide feedback on the incorrect body posture of the student through multimodal channels. The key innovations of this project are also discussed.

Replication of human body motion is a very important means to maintain a subject's emotion, knowledge and experience. The replication process requires accurate motion capturing system with sensor technologies to measure postures of human bodies and posture transmission, as well as feedback systems to adjust postures to fit into targeted ones. The sensing and feedback technologies are fundamental building blocks of motion capturing systems, work training system and rehabilitation systems. In particular, the construction of sensing and feedback systems for dynamic postures is much more complicated than that for static postures in terms of the time evolution and non-ridge body. We believe that dynamic postures can be represented by a set of blueprint or code like trajectories of particle of human body movement. Furthermore, we derive that the sensing and feedback systems should be able to establish to directly measure those critical particles without relying on external infrastructures. In the paper, some of those kinds of sensing and feedback devices are presented and some evidences such as feature contours are obtained through analysis of captured data by those devices in order to prove our estimation. We confess that our work is preliminary for this new field, but we hope that the work presented here can lead more efforts to bring out systematic approaches of feature detection and extraction of human postures whose characteristics are different from those of video and audio.

Automatic motion generation for digital character under the real-time user control is a challenging problem for computer graphic research and virtual environment applications such as on-line games. The present study introduces a methodology to generate a glove puppet animation which is controlled by a new input device, called the Smart Glove, capturing the hand motion. An animation system is proposed to generate the puppet animation based on the procedural animation and motion capture data from Smart Glove. As the control of the puppet character in the animation takes into account the design of Smart Glove and the operation of the puppet in reality, the physical hand motion can either activate the designed procedural animation through motion recognition or tune the parameters of the procedural animation to build the new puppet motion, which allows the direct user control on the animation. The potential application and improvement of the current animation system are also discussed.

This paper presents the design and validation of a wearable glove-based multi-finger motion capture device (SmartGlove) with a specific focus on the development of a new optical linear encoder (OLE). The OLE specially designed for this project has a compact size, light weight and low power consumption. The characterization tests also show that the OLE's digital output has good linearity and accuracy. The first prototype of SmartGlove which uses ten OLEs to capture the flexion/extension motion of the 14 finger joints is constructed based on the multi-point sensing method. A user study for the evaluation of SmartGlove using a standard protocol shows high repeatability and reliability in both the gripped and flat hand positions compared with four other evaluated data gloves using the same protocol.

The human body motion is a very important mean of expressing a person's emotion, knowledge and experience, as well as an effective communication tool in inter-personal interaction. We aim to provide a methodology for seamless integration of movements between the real human and the virtual one in Co-space, enabling motion replication in both directions. We developed the prototype systems consisting of a wearable InterfaceSuit that enables human motion replication and learning in Co-space, and a human-to-human motion replication methodology with multi-modal feedback mechanisms. We employed the wearable inertial sensors and optical linear encoder sensors to capture human body movement, and designed the haptic guidance device - 3 Dimensional Orientation Guide prototype with polyester tactor holder and elastic arm bands made of light-weight material for portability. We integrated sensing and feedback devices to build up a multi-resolution upper arm interfaceSuit to capture finger movements and arm movements (wrist, elbow and shoulder) and as well as wearable vibrotactile devices, sensor systems, and the control system for multimodal feedback.

Cable-driven parallel manipulators (CDPMs) are a special class of parallel manipulators that are driven by cables instead of rigid links. Due to the unilateral property of the cables, all the driving cables in a fully-constrained CDPM must always maintain positive tension. As a result, tension analysis is the most essential issue for these CDPMs. By drawing upon the mathematical theory from convex analysis, a sufficient and necessary tension-closure condition is proposed in this paper. The key point of this tension-closure condition is to construct a critical vector that must be positively expressed by the tension vectors associated with the driving cables. It has been verified that such a tension-closure condition is general enough to cater for CDPMs with different numbers of cables and DOFs. Using the tension-closure condition, a computationally efficient algorithm is developed for the tension-closure pose analysis of CDPMs, in which only a limited set of deterministic linear equation systems need to be resolved. This algorithm has been employed for the tension-closure workspace analysis of CDPMs and verified by a number of computational examples. The computational time required by the proposed algorithm is always shorter as compared to other existing algorithms.

In this paper, we present a novel sensing technique, optical linear encoder (OLE), in which the motion of an optical encoder on a reflective strip is converted to limb joints' goniometric data. A body sensing module is designed to integrate the OLE and an accelerometer. A sensor network of three sensing modules is established via Controller Areas Network (CAN) bus to capture full motion of human arm with a 7-DOF kinematic model. In addition, a statistical study was conducted to confirm the repeatability and reliability of our sensor network. Results demonstrate that the sensor system has strong potential to be used as a low-cost tool for motion capture, and objective arm function evaluation for both short-term and long-term monitoring.

This paper describe an implemented robot lion dance system, developed by a multi-disciplinary team of researchers over the past three years. We aim to use advance robotic technology to develop a mechatronic system that can perform life-size lion dancing with the traditional lion dance outfit. We intend to experiment the fusion of traditional art form and robotic technology so as to stimulate people's interest in the disappearing traditional art form and folk lords, and give new meanings to the new art form. The robot employs an event design of the robotic lion head, upper and lower body and methodology of conveying robotic lion dance into robotic lion motion. In this paper we will first describe the design concept and the architecture of the robot lion. Secondly, we will describe in detail the workings of the robotic lion dance choreography and motion control to mimic a life size lion dancing. Lastly, this paper will highlight three key challenges that were faced during the design of the robot to mimic the actual human lion dance.

This paper develops a physical model of an inertial/magnetic measurement unit by effectively integrating an accelerometer, a magnetometer, and two gyroscopes for low-g motion tracking applications. The proposed model breaks down the errors contributed by individual components, then determines error elimination methods based on sensor behavior and characteristics, and finally constructs a feedback loop for continuous self-calibration. Measurement errors are reduced by adopting a systematic design methodology: 1) tilt errors are minimized through a careful selection of A/D convertor resolution and by making compensation on sensor bias and scale factor; 2) heading errors are reduced by cancelling out nearby ferrous distortions and making tilt-compensation on the magnetometer; 3) errors from gyroscope measurements are eliminated via the least squares algorithm and continuous corrections using orientation data at the steady-state position. Preliminary tests for low-g motion sensing show that the motion tracker can achieve less than ±0.5° accuracy in tilt and less than ±1° accuracy in yaw angle measurement with above-mentioned methods.

A novel self-calibrating sensing technology using miniature linear encoders and Inertial/magnetic Measurement Unit (IMU) provides the accuracy, fast response and robustness required by many body motion processing applications. Our sensor unit consists of an accelerometer, a 3-axis magnetic sensor, 2 gyroscopes and a miniature linear encoder. The fusion of data from the sensors is accomplished by extracting the gravity related term from the accelerometer and consistently calibrating the gyroscopes and linear encoder when the sensor unit is under static conditions. Using the fused sensors, we developed a complete motion processing system that consists of a gateway where the human kinematics modeling is embedded. A time divided multiple access wireless architecture is adopted to synchronize the sensor network at 100Hz. Experimental results show that the combination of the IMU and linear encoder produces a low RMS error of 3.5° and correlation coefficient of 99.01%. A video showing the capture a performer’s upper body motion is also realized.

This paper presents the kinematic calibration of a novel 7-degree-of-freedom (DOF) cable-driven robotic arm (CDRA), aimed at improving its absolute positioning accuracy. This CDRA consists of three 'self-calibrated' cable-driven parallel mechanism (CDPM) modules. In order to account for any kinematic errors that might arise when assembling the individual CDPMs, a calibration model is formulated based on the local product-of-exponential formula and the measurement residues in the tool-tip frame poses. An iterative least-squares algorithm is employed to identify the errors in the fixed transformation frames of the sequentially assembled 'self- calibrated' CDPM modules. Both computer simulations and experimental studies were carried out to verify the robustness and effectiveness of the proposed calibration algorithm. From the experimental studies, errors in the fixed kinematic transformation frames were precisely recovered after a minimum of 15 pose measurements.

This paper presents the joint centre self-identification of a novel cable-driven anthropocentric shoulder rehabilitator. For such a wearable rehabilitator, identification of joint centre is critical for kinematic modeling, path planning and motion control. However, the shoulder joint centre location is unknown when a user wears this rehabilitator and it differs among users. These complicate the initial patient preparatory. A joint centre self-identification model is formulated based on the differential change in the cable end-point distance. A computationally effective algorithm is employed to identify the shoulder joint centre. It does not require any external measurement devices because of the redundant actuation scheme in the cable-driven shoulder rehabilitator. In order to verify the effectiveness and robustness of the proposed algorithm, we conducted computer simulation studies. In the simulation example, although large errors (up to plusmn50mm) had been injected into the initial estimates, the joint centre was always accurately identified within four to five iterations.

Workspace analysis and optimization is a critical issue in robot manipulator design. For a cable-driven parallel manipulator (CDPM), due to the unilateral driving properties of the cables, maintaining positive cable tension is essential to maneuver the moving platform. As a result, its workspaces are always determined and characterized by the tension status of its driving cables. It has been realized that the tension factor (TF) reflecting the relative tension distribution among the driving cables is an appropriate measure to evaluate the quality of force closure for CDPMs. However, in a fully restrained CDPM, since redundant cables are employed to drive the moving platform, the TF values are not unique even for a particular moving platform pose. Therefore, how to obtain the optimal TF value so as to generate an optimized workspace becomes the major subject of this paper. It is proved that the optimal TF value can be efficiently determined through a linear optimization approach, although it is essentially a nonlinear optimization problem. Subsequently, a global tension index is proposed to evaluate the quality of the entire workspace, which is achieved by integrating the local TF values (i.e., the optimal TF values at every moving platform poses) over the workspace. Computation examples are provided to demonstrate the effectiveness of the proposed algorithms

A cable-driven parallel mechanism (CDPM) possesses a number of promising advantages over the conventional rigid-link mechanisms, such as simple and light-weight mechanical structure, high-loading capacity, and large workspace. However, the formulations and results obtained for the rigid-link mechanisms cannot be directly applied to CDPMs due to the unilateral property of cables. For a CDPM, cable tensions appear to be the most important part in analyzing kinetostatic issues such as stability, workspace, stiffness, etc. This paper mainly focuses on the tension analysis issue. The equilibrium constraints such as force-closure condition, feasible wrench condition and wrench set condition are addressed. Generic analysis approaches based on dimension reduction techniques, which are successfully implemented through computationally effective recursive algorithms, are proposed.

A cable-driven parallel manipulator (CDPM) possesses a number of promising advantages over the conventional rigid-link manipulators, such as the simple and lightweight mechanical structure, high-loading capacity, and large reachable workspace. However, due to the unilateral driving capability of the flexible cables, most of the well-developed modelling and analysis methods for conventional rigid-link mechanisms cannot be directly applied to CDPM. This paper focuses on the kinematics and singularity analysis of a specific type of CDPM, i.e., a completely restrained planar CDPM. Following an analytical approach, the forward displacement analysis leads to solve a fourth order polynomial equation. Hence, the solutions can be determined efficiently in symbolic forms. To verify the effectiveness of this method, a computation example is provided, in which four real solutions are found out. It is realized the Jacobian-based singularity analysis, where all cables are treated as rigid links, is incomplete for the CDPM. Based on the concept of the instantaneous center and taking into account the cable-tension condition, a new geometrical singularity analysis method is proposed, which is able to effectively identify all forward singularity configurations for a completely restrained planar CDPM.

This paper presents a numerical technique for path planning inside the workspace of parallel manipulators avoiding singularity. A generic numerical algorithm for generating the reachable workspace of parallel manipulators is described. The singularity points are determined inside the workspace. These points are grouped into several clusters and modeled as obstacles. Subsequently, a path planning algorithm is used to find an optimal path avoiding these obstacle. If any singularity point lies on or very close to the path, the path is restructured to avoid the singularity point by a local routing method based on Grassmann's line geometry. The path planning algorithm is uniformly applicable to parallel manipulators with any combinations of revolute and prismatic joints. An example is demonstrated for the effectiveness of the algorithm.

This paper presents a numerical algorithm for the analysis of reachable workspace and singularity representation of three-legged parallel manipulators. After finding out the approximate maximum workspace, a radial and equal area discretization is done. Then, a multitasking search is performed to determine the exact workspace boundary. The volume of the workspace is determined easily by a numerical integration method. Any void inside the workspace is found out. information about its position and stretch is determined and singularity curves inside the workspace is also drawn. A three-legged modular parallel manipulator is considered to demonstrate the effectiveness of the algorithm and found to work quite satisfactorily. The model is based on Product-of-Exponential scheme of formulation and, hence, the algorithm is uniformly applicable to any combinations of revolute and prismatic joints in the configuration of the leg.

Locomotion and navigation of a surface walking/climbing robot - Planar Walker, based on a novel planar 8-bar mechanism are studied. The robot moves on a surface through decoupled transverse gaits and turning gaits with finite lengths and finite rotation angles. Motions of the gaits are modeled using planar rigid motion group. Three point-to-point navigation methods are developed for various situations: simple line of sight (SLS), simulated annealing accurate planning (SAAP), and localized hybrid accurate planning (LHAP) algorithms. Computer simulation shows that SAAP produces accurate gait sequences and LHAP saves computation time and resources for long-range targets. However, experiment shows that SLS outperforms SAAP and LHAP as the number of gaits becomes the major criteria in evaluating the gait performance due to imprecision of individual gait movements.

An instantaneous kinematics and singularity analysis of a class of three-legged, 6-DOF parallel manipulators are addressed. The kinematic relation between the actuator joint rates and end-effector velocity is established using twist annihilators. Then the singularity analysis is performed using the concepts of reciprocal screw and Grassmann line geometry. The instantaneous kinematics model derived using the product-of-exponential formulation, is uniform to any combination of revolute and prismatic joints in the leg. It is shown that reciprocal screws can be easily constructed by using twist annihilators. Based on the line geometry, the geometric conditions are proposed to identify each of the case of singularity configurations for the considered class of parallel manipulators. These geometric conditions are simple and thus, the singularity configurations are readily conceived.

The gait of a multisegment inchworm robot is a series of actuator actions that will change the shape of the robot to generate a net motion. In this article, we model the multisegment inchworm robot as a finite state automaton. Gait generation is posed as a search problem on the graph described by the automaton with prescribed state transitions. The state transitions are defined based on the kinematics of robot locomotion. The auxiliary actuator concept is introduced. Single-stride and multistride gait generations are discussed. Single-stride gaits exhibit fault-tolerant and real-time computation features that are neccessary in actual applications. Both computer simulation and experimental hardware platform are developed for various aspects of the gait generation and planning.

A modular parallel robotic system consists of a collection of individual standard units that can be assembled into various robot configurations for a diversity of task requirements. This paper is focused on the design and kinematic analysis of modular reconfigurable parallel robots. A set of fundamental modules is considered. A local frame representation of the product-of-exponentials (POE) formula, i.e., the local POE formula, is employed for the kinematic analysis of modular parallel robots. Two forward displacement analysis algorithms and a workspace visualization scheme are presented for a class of 3-leg modular parallel robots. Computation examples are also given to demonstrate the effectiveness of the proposed algorithms. The kinematic formulation shows that the local POE formula is a systematic and well-structured method for the kinematic analysis of parallel robots.

A remote arc welding robotic system has been built consisting of a six degree-of-freedom master manipulator, a stereo monitoring system, a computer control system and a slave PUMA robot. The arc welding master-slave teleoperation characteristics and the hand-eye coordination are discussed. In this paper, it is shown that the eye-in-hand monitor mode can increase operation efficiency, and that the combination of automatic weld-speed control and master-slave teleoperation mode can enhance operation stability and tracking accuracy in comparison to conventional master-slave control mode.

A remote robotic welding system has been built consisting of a six degree-of-freedom master manipulator, a stereo monitoring system, a computer control system and a slave PUMA robot. The hand-eye coordination, the human hand respondent simulation test and the influence of master unbalanced torque on arc welding master-slave teleoperation characteristics are discussed. In this paper, the power spectral density of the trajectory deviation shows a "two humps" distribution characteristic, it will also be shown that the eye-in-hand monitor mode can increase operation efficiency, and that the combination of automatic weld-speed control and master-slave teleoperation mode can enhance operation stability and tracking accuracy in comparison to conventional master-slave control mode.

Techniques are described for determining the location of the initial center of rotation (COR) of a rectangular bar being pushed or pulled. The initial COR is the point about which the bar first rotates when the pushing or pulling force is applied. This point characterizes the initial motion of the bar. Also investigated is how the location of the initial COR varies with the magnitude of the exerted force. The minimum effort criterion is proved to be able to predict the quasi-static centre of rotation. It is found that the initial COR always lies between the quasi-static and the impulsive CORs and that it will move towards the impulsive COR as the magnitude of the applied force increases. It is shown that there exists a point on an object such that, when the force is applied at that point, the object will start to rotate about a known point.>