Lux - An interactive receptionist robot for university open days

The purposes of this study are to explore the principles and rules of human-service robot division and interaction when introducing service robots into the catering service industry. Focus group interview and in-depth expert interview were conducted. The results indicated that the participants had high acceptance of importing service robot into catering service. They feel the service robots are somewhat novel and attractively. From the perspective of task orientation, the participants think that the service robots could be introduced to all of the catering service process in the future. However, they also hoped that the human and robots can cooperate and work together in the processes. They hope to have more interaction with robots. They are delight to see the service robot can bring customers different feelings and experience. They don't want to face a standard and ice-cold robot. The participants hope that the service robots have a wide variety of external appearance to meet the style of restaurants. They also anticipate the robot can act various body movements and facial expression. Immediate assistance to customers and service workers are expected. Incorrect message transmission can be tolerant by the customers, however, exact communication with service personnel is considered to be very important. The service workers have willing to work together with service robots and hope the high-risk and heavy work can be done by service robots to reduce accidents and improve the reliability and safety. The feeling of safety and more affect feeling is needed for the customers. Besides, the participants are more concerned about the value that service robots can generate. Cost down of human resources and reduced workload was expected especially. Application of service robots does not necessarily imply that manual and cognitive tasks will be replaced by machines. It is essential to serve and observe the emotional needs of customers and service workers. High-skilled workers may become more productive as they become co-workers of machines and both complement each other even at the same level of value creation. Therefore, thinking about how to cooperate and work together for the workers and service robot while introducing service robot into catering service is an important topic. The customers and service workers are looking forward to get along with an interactive, emotional, and reliable service robot.

This paper is in the context of algorithms for detection and face recognition, as well as face tracking, integrating them into service robots at home, this paper was made for service robot “Donaxi”, to the contest in RoboCup. The proposal was to set up a hybrid algorithm with different functional algorithms for the detection, recognition and face tracking, even after losing face. For detection we used the Haar algorithm, for tracking Kalman filter was used, and recognition strategy was used to segment the image into regions weighted, and discriminate against them through a decision tree for possible recognition. The experimental results show that the system is stable when saturated its field of view with many faces or persons, in this case only discriminate which is closer to the robot. It is noteworthy that followed the rules of test “Who's Who” of the category of the RoboCup at Home, this test is still an open problem for the robotics community in the area of Human-Robot Interface.

There's some very important meaning in the study of realtime face recognition and tracking system for the video monitoring and artificial vision. The current method is still very susceptible to the illumination condition, non-real time and very common to fail to track the target face especially when partly covered or moving fast. In this paper, we propose to use boosted cascade combined with skin model for face detection and then in order to recognize the candidate faces ,they will be analyzed by the hybrid Wavelet, PCA and SVM method. After that, Meanshift and Kalman filter will be invoked to track the face. The experimental results show that the algorithm has quite good performance in terms of real-time and accuracy.

Audio-visual human recognition using semi-supervised spectral learning and hidden Markov models

Face recognition systems as a biometrie system for human identification and verification has indeed shown a lot of progress over the years, illustrated through the various applications it has been used for. These applications include Human Computer Interaction systems, law, terrorist attacks, health and multimedia indexing. Furthermore, the installation of face recognition systems in public and private places helps with security improvement. However, occlusion as one of the major issues of the face recognition system tends to reduce its performance in real world applications. In this paper, the effect of occlusion scenario on face recognition technique was investigated. Also, the part of the occluded face region affecting the face recognition system is discussed.

A binocular vision system and its application to face detection and tracking in robot is introduced in this paper. With the vision system, the robot can do face detection, identification, recognition and tracking. The face area is detected in realtime by using AdaBoost algorithm. And a method is proposed with which a real face can be distinguished from a picture one by using skin color information and depth data. A specific face can be recognized by comparing the principal components of the current face to those of the known individuals in a face database built in advance. Finally, the robot can track a specified face according to depth of the face and position of a face rectangle in the frame. Experiment results are given and discussed.

Human-computer interaction for virtual-real fusion

Face recognition describes a surface framework, which is capable of processing image and detection. The proposed paper demonstrates three contributions: the first is to introduce the image representation, known as an integral image, the second application of Ada Boost learning algorithm, and the third is the cascaded framework. This includes observation, bio-metrics and video coding. Here, the primary objective is to implement a real-time system using a field-programmable gate array (FPGA) to track and detect human expression. The expression recognition involves colour-shaped coating separation and image purifying. Moreover, it involves different types of search engines which are based on computer vision. A computer controls Pan Tilt Zoom (PTZ) cameras in CCTV for surveillance systems for the face. The multi-view face tracking on a mobile platform has three methods (local binary patterns) and enhancing procedure. These face detectors are adaptable, extended for multi-opinion face recognition over the revolution ability and face matrix partition scheme is proposed to accelerate the face tracking process.

This paper presents a design and experimental study of human-robot interaction via face recognition and image tracking. A new architecture is proposed for fast face recognition of family members. In the proposed system, each family member has his/her own RBF neural networks. Each neural network is only responsible for recognizing its trained member. Consequently, the database is small and the processing time required for face recognition is minimized. A recognition rate of 94% has been achieved, an improvement relative to conventional approaches. In order to detect and track a person, we also developed an algorithm for detecting multiple faces in a scene based on division of skin and hair color regions. The face recognition and image tracking system has been integrated to an experimental mobile robot. Practical experiments reveal that the robot demonstrates real-time face recognition and tracking performance.

From the past few years, face recognition has become critical for security and surveillance applications, and is now necessary in many different settings, including offices, educational institutions, airports, corporations, and social spaces. In this paper, we present a framework of multi-face recognition for real time monitoring, resulting in simultaneous face tracking and recognition. First, the faces are detected in the video frames by using viola-jones algorithm. To remove the outliers from the detected face region, we design a face skeleton based on YCBCR color space for further feature points detection and extraction. Then harris corner feature points and SURF feature points are detected from each face, where harris points are used to track the faces in the video and the SURF feature points are used to extract facial features from the cropped faces. As the face tracking is going on, faces are simultaneously recognized by the trained classifier (support vector machine). The experiments conducted on publicly available dataset suggest that our method is reliable, accurate, and robust that can be deployed for real-world multi-face recognition systems.

Technological changes shift how visibility can be established, governed, and used. Ubiquitous visual technologies, the possibility to distribute and use images from heterogeneous sources across different social contexts and publics, and increasingly powerful facial recognition tools afford new avenues for law enforcement. Concurrently, these changes also trigger fundamental concerns about privacy violations and all-encompassing surveillance. Using the example of police investigations after the 2017 G20 summit in Hamburg, the present article provides insights into how different actors in the political and public realm in Germany deal with these potentials and tensions in handling visual data. Based on a qualitative content analysis of newspaper articles (n=42), tweets (n=267), experts’ reports (n=3), and minutes of parliamentary debates and committee hearings (n=8), this study examines how visual data were collected, analyzed, and published and how different actors legitimated and contested these practices. The findings show that combined state, corporate, and privately produced visual data and the use of facial recognition tools allowed the police to cover and track public life in large parts of the inner city of Hamburg during the summit days. Police authorities characterized visual data and algorithmic tools as objective, trustworthy, and indispensable evidence-providing tools but black-boxed the heterogeneity of sources, the analytical steps, and their potential implications. Critics, in turn, expressed concerns about infringements of civic rights, the trustworthiness of police authorities, and the extensive police surveillance capacities. Based on these findings, this article discusses three topics that remained blind spots in the debates but merit further attention in discussions on norms for visual data management and for governing visibility: (1) collective responsibilities in visibility management, (2) trust in visual data and facial recognition technologies, and (3) social consequences of encompassing visual data collection and registered faceprints.

To mitigate privacy concerns, scrambling can be used to conceal face regions present in surveillance video content. Given that lightweight scrambling tools may not protect chroma information in order to limit bit rate overhead in heterogeneous usage environments, this paper investigates how the presence of non-scrambled chroma information in face regions influences the effectiveness of automatic and human face recognition (FR). To that end, we apply three automatic FR techniques to face images that have been privacy-protected by means of a layered scrambling technique developed for Motion JPEG XR, testing the effectiveness of automatic FR and layered scrambling using various experimental conditions. In addition, we investigate whether agreement exists between the judgments of 32 human observers and the output of automatic FR. Our experimental results demonstrate that human observers are not able to successfully recognize face images when simultaneously visualizing scrambled luma and non-scrambled chroma information. However, when an adversary has access to the coded bit stream structure, the presence of non-scrambled chroma information may significantly contribute to privacy leakage. By additionally applying layered scrambling to chroma information, our experimental results show that the amount of privacy leakage can be substantially decreased at the cost of an increase in bit rate overhead, and with the increase in bit rate overhead dependent on the number of scalability layers used.

We propose the affine filter bank modulation (AFBM) waveform for enhanced integrated sensing and communications (ISAC) in sixth generation (6G), designed by drawing on concepts from classical filter bank multicarrier modulation (FBMC) theory and recent advances in chirp-domain waveforms, particularly affine frequency division multiplexing (AFDM). Specifically, AFBM exhibits several desirable properties, with emphasis on its remarkably low peak-to-average power ratio (PAPR) and reduced out-of-band emission (OOBE) when bench-marked against the conventional AFDM waveform under doubly-dispersive (DD) channel conditions. In the communications setting, reliable symbol detection is achieved using a tailored low-complexity Gaussian belief propagation (GaBP)-based algorithm, while in the sensing setting, a range and velocity estimation approach is developed that integrates an expectation maximization (EM)-assisted probabilistic data association (PDA) framework to accurately identify surrounding targets. The highlighted performance and benefits of AFBM are validated through analytical and numerical evaluations, including conventional metrics such as ambiguity function (AF), bit error rate (BER), and root mean square error (RMSE), consolidating its position as a promising waveform for next-generation wireless systems.

Small informal meetings of two to four participants are very common in work environments. For this reason, a convenient way for recording and archiving these meetings is of great interest. In order to efficiently archive such meetings, an important task to address is to keep trace of “who talked when” during a meeting. This paper proposes a new multi-modal approach to tackle this speaker activity detection problem. One of the novelty of the proposed approach is that it uses a human tracker that relies on scanning laser range finders (LRFs) to localize the participants. This choice is especially relevant for robotic applications as robots are often equipped with LRFs for navigation purpose. In the proposed system, a table top microphone array in the center of the meeting room acquires the audio data while the LRF based human tracker monitors the movement of the participants. Then the speaker activity detection is performed using Gaussian mixture models that were trained before hand. An experiment reproducing a meeting configuration demonstrates the performance of the system for speaker activity detection. In particular, the proposed hands free system maintains an good level of performance compared to the use of close talking microphone while participants are simultaneously speaking.

In this paper, single-target tracking using radar measurements is addressed. Recently, algorithms based on Bernoulli random finite sets have proved efficient in a cluttered environment. However, in Bayesian approaches, the choice of the motion model impacts the trajectory estimation accuracy. To select an appropriate set of motion models, a joint tracking and classification (JTC) algorithm can be used. The principle is to consider different target classes depending on their maneuvrability, each of them being associated to a set of motion models. In this context, additional information such as a target length extent measurement can improve both classification and trajectory estimation. Therefore, we propose a multiple-model Bernoulli filter to perform JTC. To jointly estimate the trajectory and the target length which is constant, a Rao-Blackwellized approach is considered. Another contribution is that a bank of probabilistic data association filters is run instead of Kalman filters to account for false detections.