Search

Tidal observations influence the transport efficiency of international commercial ports and can be affected by mechanical failures or typhoon-induced storms. These factors cause observational interruptions, which lead to tidal data loss or anomaly. Thus, the applicability of the data is reduced. Existing methods still have certain limitations in accurately predicting the tide level because of the omission of a large amount of data. Therefore, missing value imputation and tide level forecasting of tidal data are crucial topics in tidal observation study. In this study, we propose a deep learning algorithm for missing value imputation and tide level forecasting of tidal data. The test data are obtained from the tidal data of Keelung Port, Taipei Port, Tamsui Port, Taichung Port, Jiangjun Port, Anping Port, Kaohsiung Port, Hualian Port, Suao Port, and Penghu Port, constructed by the Harbor and Marine Technology Center, Taiwan. The average error value for conducting missing value imputation is 0.086 m ± 5%, and the average error value for tide level forecasting is 0.071 m. The experimental results reveal that the deep neural network has better performance than the traditional statistical methods and other artificial neural networks.

As robots are put into humans’ daily life, the assigned tasks to robots are varied, and the different needs of people interacting with robots are immense. As a result, when facing different users, it is important for robots to personalize the interactions and provide user-desired services. This paper, therefore, proposes a learning strategy on the service-providing model. Through human feedback, the strategy enables the robot to learn the users’ needs, as well as preferences, and adjust its behaviors. Here, we assume that users’ needs and preferences may vary with time; hence the goal of this paper is to let the adjustment of robot behaviors be able to adapt to those variations. In turn, the service-providing model of the robot could adjust online as well. That is, it can select a new action from those favorable actions that have already been selected or an action that is not an unfavorable action but has annoyed humans recently. To implement our system, the service robot under discussion is applied to the home environment. For performance evaluation, we have performed extensive experiments that satisfactorily demonstrate that our robot can provide services to different users and adapt to their preference change.

Trust is a key component in developing successful interpersonal relationships. In this paper, we posit that the same is true for Human-Robot Interaction (HRI), since human trust toward robots can facilitate HRI in terms of comfort and usability. We investigated the ability of a socially assistive robot to promote trust in the social relationship with its user by inducing self-disclosure of the user’s negative experiences and offering coping mechanisms to deal with these. To achieve this purpose, our system is equipped with deep learning techniques to detect the user’s negative facial expressions, which in turn can be used as cues for the robot to proactively induce self-disclosure. Once triggered, using a conversational model, the robot engages the user to determine the cause of their negative mood. Then, it infers the user’s internal feelings by applying Markov Chain Monte Carlo (MCMC) inference over a Bayesian Network on the user’s utterance. Combining the information gathered from the concept inferencing process and the self-disclosure content, the system is able to estimate a set of desires from the Bayesian Network. Experiments show that our proposed work can correctly infer the user’s feelings and desires from their utterances, as well as generate an appropriate response, resulting in the improvement of human’s trust toward the robot.

In a situation where a robot initiates conversation with a group of people, questions such as "where is the people group?" and "whether the robot should approach them?" should be addressed. This paper develops a new system that enables a robot to determine whether or not it should approach the aforementioned human group and interact with them after identifying what the current social situation is. The system is mainly to fuse depth-related data to track the positions of a group of people, extract social cues of those people by using depth-related data and a decision network (DN) model, and the main challenge lies in understanding the social cues of the group and the current underlying social situation concerning the relation between the robot and the group. The social cues are based on Proxemics and F-formations, whereas the social situations are categorized as individual-to-individual, individual-to-robot, robot-to-individual, group-to-robot, robot-to-group, confidential discussion and group discussion. Our system proceeds as follows : once a group of people are detected and the social cues of that target group of people are extracted, the corresponding social situation is appropriately inferred, and in turn the robot decides whether it should initiate conversation with the group based on rules to be specified later. The conducted experimental results demonstrate the properness of the system design and the efficacy of the proposed method in recognizing the social cues among individuals of the group as well as the nature of the social situations concerning the group and the robot. We developed a new system to allow a robot to determine whether it should approach a human group and interact with them by inferring what the current social situation is.The proposed system tracks the positions of people through the fusion of different depth-related data within a long range and complex environment.Social situations concerning the robot and the target group of people are inferred first and then the robot decides whether it should initiate interactions or not in the human-robot social domain.

In this paper, we propose a sensory cues guided robotic walker for improving gaits of Parkinson Disease (PD) patients. A completely non-intrusive, 3D real-time leg pose tracking and gait analysis are proposed by using a depth camera mounted on the rear of the robotic walker. It has been studied that the sensory cues can serve as effective stimuli to the PD patients for gait improvement. In our work, the sensory cues including visual and auditory cues are incorporated into the robotic walker. More specifically, both sensory cues are gait-adaptive, of which the visual cue in particular is projected onto the ground by a projector installed on the walker, in order to stimulate patients walking gait more easily. Since the adjustable cues can improve patients gaits and reduce their uncomfortableness simultaneously, the hereby developed robotic walker serves as a gait rehabilitation mechanism. To demonstrate the performance of the developed walker, several real experiments have been conducted. First, the accuracy of the proposed 3D leg pose tracking is verified by a standard motion capture system. Next, seven participants (4PD patients and 3 healthy elders) are recruited to test the system three days for verifying the effectiveness of participants gait improvement. The experimental results confirm the potential of the walker serving as a rehabilitation device for PD patients.

For a service robot, it is not adequate to let its navigational movement be based only on a single metric, such as minimum distance path. In the environment where the robot and humans are coexisting, the robot should always perform social navigation whenever it is moving. However, to perform social navigation, the robot needs to follow certain “social norms” of the environment. Recently, deep reinforcement learning (DRL) technique is popularly applied to the robotics field; yet, it is rarely used to solve the mentioned social navigation problem, generally deemed as a high dimension complex problem. In this paper, we propose the composite reinforcement learning (CRL) framework under which the robot learns appropriate social navigation with sensor input and reward update based on human feedback. For learning the aspect of human robot interaction (HRI), we provide a method to facilitate the training of DRL in real environment by incorporating prior knowledge to the system. It turns out that our CRL system not only can incrementally learn how to set its velocity and to perform HRI but also keep collecting human feedback to synchronize the reward functions to the current social norms. The experiments show that the proposed CRL system can safely learn how to navigate in the environment and show that our system is able to perform HRI for social navigation.

For a visual servo system, there usually exists the problem of time-delay likely caused by long image processing and data transmission. The visual servo system of our robot is subject to two main limitations stemming from the specific commercial mobile manipulator, of which one is the large time-delay due to image transmission, whereas the other is failure to directly command each joint velocity as the visual servo's control input. In this paper, we propose a novel visual servo system to compensate the large time-delay, and we use integral of velocity which incorporating the image estimation and prediction with the kinematic model so as to achieve indirect velocity control with position based control mode joint. Our proposed visual servo system controls the neck joints of robot and keeps the object within the field of view of the camera during the approaching phase. We evaluate the proposed approach by experiment on a real wheeled humanoid robot and find the resulting performance quite promising, showing that our method is feasible and practical.

This paper presents a hybrid system of applying homeostatic drives theory and stimuli from environment or human on a robot, aiming to solve the increasing need of elder caring in modern aging society. Through reinforcement learning with human feedback, the robot is enabled to make appropriate decisions by reviewing its own motivation, environment stimuli and human inputs, allowing the robot to finish vague tasks, i.e. to serve, help or have a conversation with human on different situations. In our experiment, the human satisfaction rate could reach 94% and the robot could as well maintain a stable condition to perform long-term service. If the robot is to be deployed in a nursing home, it can further serve and meet the requirements of elders, while simultaneously maintaining its own system, such as battery level, in regular daily lives.

To enable service robots enter a multi-human office environment, it is important to find a group of human users’ social patterns and then to provide a proper service to them in time. Usually, human users’ social patterns are represented in terms of nonverbal social signals. In this paper, some nonverbal social signals are fast detected in social environments. Then, robot can find the spatial social patterns of human users. Those patterns are indicated to human-to-human, human-to-robot or multi-human-to-robot interaction. Experimental results shows that our robot successfully find the aforementioned users’ social patterns.

In this paper, we propose a framework to perceive the level of intimacy in dyadic human interactions from both robot perspective and first-person perspective. First of all, for catching the insight of human interaction with three different degrees of intimacy persons, namely normal, familiar and close, we have done a preliminary user study of social interaction. Next, from the field of social science and our study, we design four types of social interaction features, consisting of proxemics, non-verbal, verbal, and temporal features to categorize the intimacy level from observations, but including three of them reaches the best result. Finally, to validate our work here, several experiments have been conducted, and the results show that the framework perceives the aforementioned intimacy level with accuracy up to 86.11% in average.

People engaging in an activity usually has individual tolerance to be interrupted [1], [2]. Humans subconsciously adapt their behaviors to draw other one's attention and to get into a conversation based on their historical experiences, but robots often fail to be aware of humans' feeling and thus interrupt their users repeatedly. To endow service robots with such socially acceptable ability, we propose an online human-aware interactive learning framework in this paper, under which the robot personalizes its behaviors according to both observed user's attention and its conjecture about user's awareness of itself. To this purpose, the correlation between the robot's theory of awareness, user's attention and robot behavior are explored through reinforcement learning techniques. The conducted experiment shows that the robot can personalize its interruption strategy, and the optimal policies converged for at least 26 episodes.

In this paper, we develop a human awareness Decision Network model for robot performance on decision making. To accomplish more natural and intelligent human robot interaction (HRI), a robot should not only be able to infer the user's intention through recognizing the actions, but also to perform appropriate decisions and to learn from the user's feedback. In traditional approaches, user intention inference and feedback learning are dealt with separately. In this paper, we propose an integrated strategy of human-oriented perception, user modeling and user sensitivity in a social environment. The robot can analyze a user's feedback to adjust its decisions as the user expects through the strategy. The experimental results show the effectiveness of the proposed approach that enables autonomous adaptation of robot's decision to the user desires. Also, we demonstrate a satisfactory performance in terms of successful inference of human intentions, as well as adequacy of the decisions made by the robot for meeting user expectation.

In this paper, we propose an online scene reconstruction algorithm with monocular camera since there are many advantages on modeling and visualization of an environment with physical scene reconstruction instead of resorting to sparse 3D points. The goal of this algorithm is to simultaneously track the camera position and map the 3D environment, which is close to the spirit of visual SLAM. There're plenty of visual SLAM algorithms in the current literature which can provide a high accuracy performance, but many of them rely on stereo cameras. It's true that we'll face many more challenges to accomplish this task with monocular camera. However, the advantages of cheaper and easier deployable hardware setting have made monocular approach more attractive. Specifically, we apply a maximum a posteriori Bayesian approach with optimization technique to simultaneously track the camera and build a dense point cloud. We also propose a feature expansion method to expand the density of points, and then online reconstruct the scene with a delayed approach. Furthermore, we utilize the reconstructed model to accomplish visual localization task without extracting the features. Finally, a number of experiments have been conducted to validate our proposed approach, and promising performance can be observed.

In this paper, we present a novel data structure representing the environment with occupancy grid cells while each grid map is associated with a set of line features extracted from laser scan points. Due to the fact that line segments are principal elements of artificial environments, they provide considerable geometric information about the environment which can be used for enhancing the accuracy of localization. Orthogonal characteristic of line features is the key issue to guarantee the consistency of the SLAM algorithm by allowing us to deal with lines that are parallel or perpendicular to each other. This behavior allows us to sample robot poses more correctly. As a result, the proposed algorithm can close bigger loops with the same number of particles. Experimental results are carried out using SICK LMS-100 laser scanner which has a maximum range of 20m and Pioneer 3DX mobile robot mapping an indoor environment with the size of 40m × 47m.