Search

Exploiting hand gestures for non-verbal communication has extraordinary potential in HCI. A data glove is an apparatus widely used to recognize hand gestures. To improve the functionality of the data glove, a highly stretchable and reliable signal-to-noise ratio sensor is indispensable. To do this, the study focused on the development of soft silicone microchannel sensors using a Eutectic Gallium-Indium (EGaIn) liquid metal alloy and a hand gesture recognition system via the proposed data glove using the soft sensor. The EGaIn-silicone sensor was uniquely designed to include two sensing channels to monitor the finger joint movements and to facilitate the EGaIn alloy injection into the meander-type microchannels. We recruited 15 participants to collect hand gesture dataset investigating 12 static hand gestures. The dataset was exploited to estimate the performance of the proposed data glove in hand gesture recognition. Additionally, six traditional classification algorithms were studied. From the results, a random forest shows the highest classification accuracy of 97.3% and a linear discriminant analysis shows the lowest accuracy of 87.4%. The non-linearity of the proposed sensor deteriorated the accuracy of LDA, however, the other classifiers adequately overcame it and performed high accuracies (&gt, 90%).

Because the internal structure, degree of freedom, skin control position and range of the android face are different, it is very difficult to generate facial expressions by applying existing facial expression generation methods. In addition, facial expressions differ among robots because they are designed subjectively. To address these problems, we developed a system that can automatically generate robot facial expressions by combining an android, a recognizer capable of classifying facial expressions and a genetic algorithm. We have developed two types (older men and young women) of android face robots that can simulate human skin movements. We selected 16 control positions to generate the facial expressions of these robots. The expressions were generated by combining the displacements of 16 motors. A chromosome comprising 16 genes (motor displacements) was generated by applying real-coded genetic algorithms, subsequently, it was used to generate robot facial expressions. To determine the fitness of the generated facial expressions, expression intensity was evaluated through a facial expression recognizer. The proposed system was used to generate six facial expressions (angry, disgust, fear, happy, sad, surprised), the results confirmed that they were more appropriate than manually generated facial expressions.

This paper proposes a soft linear actuator imitating a foxtail movement and introduces two applications. The proposed biomimetic soft linear actuator mainly consists of three parts: a soft core containing a vibration motor with silicone spikes on its surface, a tubular passage to guide the movement of the soft core, and a switching mechanism to change the direction of the movement of the soft core when an excessive tensile force is applied. When vibration is generated, the soft core moves along the tubular passage toward one direction depended on the leaning direction of the spikes; namely, it can be utilized as a linear actuator to manipulate an object if the core is wired onto it. By inserting a spring at the end of the wire, the leaning direction of the spikes can be changed reversely if the tensile force between the core and the object connected by wire-spring exceeds either the kinetic force toward movement direction generated by the core or the frictional force between the spikes and the tubular passage. Consequently, this characteristic endows the proposed soft linear actuator with a gentle and safe manipulation to handle the object. We present the two applications of gentle manipulation: an elbow manipulator mimicking a human bicep to lift its lower arm and a five-finger robot hand grasping a fragile object. By testing these two applications, we could substantiate the feasibility of the proposed soft linear actuator as a core element to realize safe human-robot interaction or cooperation.

This paper presents the bio-mimetic design approach, the dynamic model, and potential applications for a hybrid soft actuator. The proposed hybrid soft actuator consists of two main parts: a cylinder-shaped rigid core and soft silicone spikes wrapped around the core’s surface. The key idea of the proposed design approach is to mimic the movement of a grass-spike at a functional level by converting the vibration force generated by a small electric motor with a counterweight in the rigid core into a propulsion force produced by the elastic restoration of the spikes. One advantage of this design approach is that the hybrid soft actuator does not need to be tethered by a tube line from an air compressor and is more amenable to fine control. In addition, the hybrid soft actuator can be modularized with a wire and a tubular passage, which in turn work as a linear actuator. The dynamic model of the hybrid soft actuator can be derived by applying Lagrangian mechanics, and unknown system parameters can be identified by the optimization process based on the empirical data. Two applications—an elbow manipulator and a robotic hand grasper—demonstrate the feasibility of the proposed actuator to perform a muscle-tendon action successfully.

Android robots for arts performances are required to perform human-like actions. Recently, the most popular method is to reconstruct the action of the robot by video-recording the choreographer's action directed by a performance director in a studio with 3D motion capture equipment. In this paper, we propose a method to create robot motion data based on video-recorded human demonstration. The proposed method aims at the following two objectives: 1) the robot motion can be made through the existing performance video clips demonstrated by a human actor, and 2) the partial corrections for the robot motion can be made instantaneously when the corrections are given by performance director. To achieve those two objectives, we first adopted OpenPose to extract the coordinates of human joints from a video clip of human performances. Then, the extracted coordinates were converted into the joint angle values. Next, the joint angle values were renovated to generate robot motions satisfying the kinematics and hardware specifications of the robot. Finally, the efficacy of the proposed method is examined by applying the generated robot motions to a female android robot.

Facial expressions are a nonverbal way to express our intent or feelings to others. Human-like android robots should also be able to make facial expressions for people to express natural behavior. We should generate the natural facial expressions of the android robot, but there is no rule for expressing it naturally. So, we have created facial expressions by controlling the motors one by one. This is a very repetitive and inefficient procedure, and the resulting expressions will vary depending on the developer's skill. In this paper, we propose a method to generate facial expressions based on facial action unit defined by Ekman. We used OpenFace to recognize the action unit of the face robot and obtained 13 recognizable action units. The maximum likelihood estimation was used to generate a probabilistic model of the motor parameter values representing the action unit and the least-square method was used to classify the motors involved in the action unit expression.

Cutaneous sensory feedback can be used to provide additional sensory cues to a person performing a motor task where vision is a dominant feedback signal. A haptic joystick has been widely used to guide a user by providing force feedback. However, the benefit of providing force feedback is still debatable due to performance dependency on factors such as the user’s skill-level, task difficulty. Meanwhile, recent studies have shown the feasibility of improving a motor task performance by providing skin-stretch feedback. Therefore, a combination of two aforementioned feedback types is deemed to be promising to promote synergistic effects to consistently improve the person’s motor performance. In this study, we aimed at identifying the effect of the combined haptic and skin-stretch feedbacks on the aged person’s driving motor performance. For the experiment, fifteen healthy elderly subjects (age 72.8±6.6 years) were recruited and were instructed to drive a virtual power-wheelchair through four different courses with obstacles. Four augmented sensory feedback conditions were tested: no-feedback, force feedback, skin-stretch feedback, and a combination of both force and skin-stretch feedbacks. While the haptic force was provided to the hand by the joystick, the skin-stretch was provided to the steering forearm by a custom-designed wearable skin-stretch device. We tested two hypotheses: i) an elderly individual’s motor control would benefit from receiving information about a desired trajectory from multiple sensory feedback sources, and ii) the benefit does not depend on task difficulty. Various metrics related to skills and safety were used to evaluate the control performance. Repeated measure ANOVA was performed for those metrics with two factors: task scenario and the type of the augmented sensory feedback. The results revealed that elderly subjects’ control performance significantly improved when the combined feedback of both haptic force and skin-stertch feedback was applied. The proposed approach suggest the feasibility to improve people’s task performance by the synergistic effects of multiple augmented sensory feedback modalities.

In the field of human rehabilitation, user satisfaction is a vital criterion in assessing the efficacy of assistive devices. A wheelchair is a popularly researched assistive device with more than 3.5 million users in the US [1]. Through subsequent innovations, the beginnings of the wheelchair have undergone tremendous transformations to result in today's powered-wheelchair. This technological advancement has been bolstered with studies that prove the powered-wheelchair's positive impact on user's mentality and on the social cost savings per user [2].

Falling accidents are costly due to their prevalence in the workplace. Slipping has been known to be the main cause of falling. Understanding the motor response used to regain balance after slipping is crucial to developing intervention strategies for effective recovery. Interestingly, studies on spinalized animals and studies on animals subjected to electrical microstimulation have provided major evidence that the Central Nervous System (CNS) uses motor primitives, such as muscle synergies, to control motor tasks. Muscle synergies are thought to be a critical mechanism used by the CNS to control complex motor tasks by reducing the dimensional complexity of the system. Even though synergies have demonstrated potential for indicating how the body responds to balance perturbations by accounting for majority of the data set's variability, this concept has not been applied to slipping. To address this gap, data from 11 healthy young adults were collected and analyzed during both unperturbed walking and slipping. Applying an iterative non-negative matrix decomposition technique, four muscle synergies and the corresponding time-series activation coefficients were extracted. The synergies and the activation coefficients were then compared between baseline walking and slipping to determine shared vs. task-specific synergies. Correlation analyses found that among four synergies, two synergies were shared between normal walking and slipping. However, the other two synergies were task-specific. Both limbs were contributing to each of the four synergies, suggesting substantial inter-limb coordination during gait and slip. These findings stay consistent with previous unilateral studies that reported similar synergies between unperturbed and perturbed walking. Activation coefficients corresponding to the two shared synergies were similar between normal walking and slipping for the first 200 ms after heel contact and differed later in stance, suggesting the activation of muscle synergies in response to a slip. A muscle synergy approach would reveal the used sub-tasks during slipping, facilitating identification of impaired sub-tasks, and potentially leading to a purposeful rehabilitation based on damaged sub-functions.

This paper presents the strategy for searching a hidden object in an unknown area for using by multiple distributed autonomous robotic systems (DARS). To search the target in Markovian space, DARS should recognize their surrounding at where they are located and generate some rules to act upon by themselves. First of all, DARS obtain 6-distances from itself to environment by infrared sensor which are hexagonally allocated around itself. Second, it calculates 6-areas with those distances then take an action, i.e., turn and move toward where the widest space will be guaranteed. After the action is taken, the value of Q will be updated by relative formula at the state. We set up an experimental environment with five small mobile robots, obstacles, and a target object, and tried to research for a target object while navigating in a un known hallway where some obstacles were placed. In the end of this paper, we present the results of three algorithms - a random search, an area-based action making process to determine the next action of the robot and hexagon-based Q-learning to enhance the area-based action making process.

In this paper, we present the area-based Q-learning to search a target object using multiple robot. To search the target in Markovian space, the robots should recognize their surrounding at where they are located and generate some rules to act upon by themselves. Under area-based Q-learning, a robot, first of all, obtains 6-distances from itself to environment by infrared sensor which are hexagonally allocated around itself. Second, it calculates 6-areas with those distances then take an action, i.e., turn and move toward where the widest space will be guaranteed. After the action is taken, the value of Q will be updated by relative formula at the state. We set up an experimental environment with five small mobile robots, obstacles, and a target object, and tried to search for a target object while navigating in a unknown hallway where some obstacles were placed. In the end of this paper, we presents the results of three algorithms - a random search, area-based action making (ABAM), and hexagonal area-based Q-teaming.

Generally, genetic algorithm (GA) has too much time and space complexity when it is running in the typical processor. Therefore, we are forced to use the high-performance and expensive processor by this reason. It also works as a barrier to implement real device, such a small mobile robot, which is required only simple rules. To solve this problem, this paper presents and proposes enhanced processor-architecture for the faster GA processing. A typical processor architecture can be enhanced and specialized by two approaches: one is a sorting network, the other is a residue number system (RNS). A sorting network can improve the time complexity of which needs to compare the populations` fitness. An RNS can reduce the magnitude of the largest bit that dictates the speed of arithmetic operation. Consequently, it can make the total logic size smaller and innovate arithmetic operation speed faster.

This paper presents and proposes two approaches to speed up for the controller unit. One is a residue number system (RNS), the other is a sorting network. Using an RNS reduces the total bit size of adder and carry-propagation delays. Also, using sorting network makes comparing time of 16 input data sequences within 10 clocks. First, this paper introduces the basic concept of an RNS and a sorting network. Next, we design the controller using the RNS adder and Green model 16-input sorting network. We present experimental results about elevator group control by applying these two approaches. Experimental results show that the RNS can reduce adder size, which calculates an elevator attractive function (EAF) value, to 5-bit and its carry propagation delay within a single residue. The results also demonstrate that a sorting network is able to choose the best elevator to serve against passenger's request by comparing the attractive function value of each elevator with in 4 clocks. It means that our primary goal can be achieved by these two approaches.

This paper presents the hexagon-based Q-leaning for object search with multiple robots. We set up an experimental environment with five small mobile robots, obstacles, and a target object. The robots were out to search for a target object while navigating in a hallway where some obstacles were placed. In this experiment, we used three control algorithms: a random search, an area-based action making (ABAM) process to determine the next action of the robots, and hexagon-based Q-learning to enhance the area-based action making process.