Search

SUMMARYThis article proposes a method for incremental data dimensionality reduction in loop closure detection for robotic autonomous navigation. The approach uses dominant eigenvector concept for: (a) spectral description of visual datasets and (b) representation in low dimension. Unlike most other papers on data dimensionality reduction (which is done in batch mode), our method combines a sliding window technique and coordinate transformation to achieve dimensionality reduction in incremental data. Experiments in both simulated and real scenarios were performed and the results are suitable.

This article describes the elaboration of a low-cost system that allows user interaction with VR environments modeled in Unity and viewed through Google Cardboard. The proposed system divides the processing in a client-server architecture in which the VR experience is executed on the client, while it sends pictures captured from the mobile device inside the Card Board to the server. This server processes the hand pose information and sends it to the client.

Landing Signalman (LS) is the military in charge of visually signaling the helicopter pilot when landing and taking off from a moving ship, ensuring general safety conditions of the flight deck area. It requires self-confidence, knowledge, skills, team coordination, and appropriate reaction capacity only achieved with intensive training. After studying the theory, Brazilian Navy LS trainees go directly to the practice stage at sea. This hands-on training phase straight after classroom reveals a series of limitations on LS trainee’s performance. To better prepare the LS trainee, reduce training costs and provide a safe environment to perform unlimited training opportunities, we developed a lightweight, portable, and low-cost VR simulator for training optimization, denominated Helicopter Visual Signal Simulator (HVSS). Due to its specificities, a Machine Learning gesture recognition method was integrated into the virtual environment, and two datasets were created. This study details the process of achieving the design goals, developing the VR simulator prototype, and analyzing the tests we performed With 15 experienced instructors. The results indicate that HVSS meets all requirements to provide a reliable training solution.

In this article, we extend our previous instrument [1], which helps trainees improve the precision of their compressions during Cardio-Pulmonary Resuscitation (CPR), by adding to it a 3D-printed component (http://acima.lncc. br/i&m) that further improves the precision of force angle measurements. Our instrument measures the force vector and frequency of CPR compressions performed by the trainee. CPR is a technique that allows a certified person to try to keep alive someone whose heartbeat and/or breathing has stopped. Through compressions applied to the thorax of the individual in need of help, one can maintain the blood flow and air intake until further help (and equipment) arrives. During CPR training, our 3D-prototype methodically attaches a standard CPR manikin to a Nintendo WiiTM Balance Board, enabling adequate balance of the instrument, and thus, obtaining more precise data out of the compressions performed. Results of our empirical evaluations show that with the 3D prototype, the measured parameters of force value, force direction and frequency have a standard uncertainty of 0.91 KgF, 0.25 Rad (14.35 degrees), and 0.15 Hz, respectively, while without our 3D prototype they have a standard uncertainty of 0.72 KgF, 0.44 Rad (25.25 degrees) and 0.12 Hz, respectively. Results prove that our 3D prototype contributes to increasing the precision of force angle measurement, since it holds the manikin firmly in place, while having negligible effect in force amount and frequency measurements.

Cardiac catheterism is important because it offers many advantages in comparison to open surgery, for example, fewer injuries, lower risk of infections, and shorter recovery times. Simulators play a fundamental role in training packages, and virtual learning environments are less stressful. Moreover, they can also be used in certification boards and in performance assessments. A realistic and interactive simulator must be fast. In this work, the physical model of the guidewire used in catheter simulations has been improved. In particular, we determined a simple analytic expression to calculate the direction of a guidewire segment, which minimizes the total energy. The surface energy resulting from the guidewire--artery interaction and the bending energy of the guidewire is approximated up to the second order, which gives rise to interactions between segments. Furthermore, the multiple segment relaxations are introduced, enhancing the convergence especially at the beginning of the relaxation cycle. The formulas are written in matrix form of dimension 4M×4M, where M represents the number of segments varied in the update step. The method results in a more stable static solution.

The practice of observation in Army Artillery has been hampered by the high cost and the limited number of available instruction areas. Using Virtual Reality (VR) becomes a viable alternative for this important training activity. In this context, was developed the first Brazilian Army VR simulator in which an instructor can train an observer in a virtual environment of Artillery employment. This simulator was tested by a sample of 13 Artillery officers with extensive practical experience in observation. The results were compared with reference values obtained from related work. The data show that the simulator reached the reference values for the two parameters evaluated: presence and effectiveness.

An eating disorder is a disturb characterized by an atypical eating behavior. The most common strategy for treatment is CBT, and its limitations have shown to be overcome using virtual reality approaches. Most VR envi- ronments created for this purpose reproduce tasks for treatment of body image disturbances and food avoidance. This work presents a new approach for eating disorders treatment using VR and concepts of serious games and gamification, which have been shown of promising potential in health context due to its ad- vantages when compared to other approaches. Tests assessing the usability and sense of presence in the virtual environment were applied to a group of 10 par- ticipants, indicating that the environment is well suited for medical treatment.

This paper presents a real-time framework that combines depth data and infrared laser speckle pattern (ILSP) images, captured from a Kinect device, for static hand gesture recognition to interact with CAVE applications. At the startup of the system, background removal and hand position detection are performed using only the depth map. After that, tracking is started using the hand positions of the previous frames in order to seek for the hand centroid of the current one. The obtained point is used as a seed for a region growing algorithm to perform hand segmentation in the depth map. The result is a mask that will be used for hand segmentation in the ILSP frame sequence. Next, we apply motion restrictions for gesture spotting in order to mark each image as a `Gesture' or `Non-Gesture'. The ILSP counterparts of the frames labeled as "Gesture" are enhanced by using mask subtraction, contrast stretching, median filter, and histogram equalization. The result is used as the input for the feature extraction using a scale invariant feature transform algorithm (SIFT), bag-of-visual-words construction and classification through a multi-class support vector machine (SVM) classifier. Finally, we build a grammar based on the hand gesture classes to convert the classification results in control commands for the CAVE application. The performed tests and comparisons show that the implemented plugin is an efficient solution. We achieve state-of-the-art recognition accuracy as well as efficient object manipulation in a virtual scene visualized in the CAVE.

The accuracy of the results obtained with Finite Element Methods depends on the basis functions employed to approximate the displacement fields. If the beam is very thin, there can be shear locking and the numerical approximation leads to erroneous solutions. In this work, shear locking is analyzed by calculating expressly the curves of the transverse deflection, the cross-section rotation, and the shear strain in different approaches. The influence of the ratio between the shear stiffness and the bending stiffness is explicitly shown when force and moment loads are applied. It is concluded that the locking behavior at nodes is not the decisive factor to assess the quality of the solution. The important point is to analyze the entire curves when the ratio between the stiffnesses is varied. It is verified that the mixed interpolation and the discrete shear gap approaches are superior to the pure displacement and to the field consistency approaches.

Cardio-Pulmonary Resuscitation (CPR) is a technique that allows a CPR certified person to keep alive someone whose heartbeat and/or breathing has stopped. Through compressions applied to the thorax of the individual in need of help, one can maintain the blood flow and air intake until further help (and equipment) arrives. Training for CPR is performed using manikins that mimic a human torso. In this paper, we design an instrument for measuring the force vector (magnitude and direction) and frequency of CPR compressions applied during CPR training sessions, to help a trainer objectively evaluate the performance of trainees. Excessive/insufficient force or frequency of compressions will harm, not help, the person in need. So it is important for a trainer to measure these parameters. Our instrument uses the standard CPR manikin plus a Nintendo Wii Fit balance board, and is therefore practical and affordable. Our experiments conducted with real subjects show that our system can measure these parameters with a standard uncertainty of 1.96 N, 0.79 degrees, and 0.26 Hz, respectively for force, direction and frequency.

In the present work, a simulation system is proposed that can be used as an educational tool by physicians in training basic skills of minimally invasive vascular interventions. In order to accomplish this objective, initially the physical model of the wire proposed by Konings has been improved. As a result, a simpler and more stable method was obtained to calculate the equilibrium configuration of the wire. In addition, a geometrical method is developed to perform relaxations. It is particularly useful when the wire is hindered in the physical method because of the boundary conditions. Then a recipe is given to merge the physical and the geometrical methods, resulting in efficient relaxations. Moreover, tests have shown that the shape of the virtual wire agrees with the experiment. The proposed algorithm allows real-time executions, and furthermore, the hardware to assemble the simulator has a low cost.

Foreword to the Special Issue of SVR 2017

We propose a framework with a flexible architecture that have been designed and implemented for collaborative interaction of users, to be applied in massive applications through the Web. We introduce the concept of interperception and use technologies as massive virtual environments and teleoperation for the creation of environments (mixing virtual and real ones) in order to promote accessibility and transparency in the interaction between people, and between people and animate devices (such as robots) through the Web. Experiments with massive games, with interactive applications in digital television, with users and robots interacting in virtual and real versions of museums and cultural centers are presented to validate our proposal.

This work presents the design and the features of a flexible realtime 3D graphics engine aimed at the development of multimedia applications and collaborative virtual environments. The engine, called EnCIMA (Engine for Collaborative and Immersive Multimedia Applications), enables a fast development process of applications by providing a high level interface, which has been implemented using the C++ object-oriented programming paradigm. The main features of the proposed engine are the support to scene management, ability to load static and animated 3D models, particle system effects, network connection management to support collaboration, and collision detection. In addition, the engine supports several specialized interaction devices such as 3D mice, haptic devices, 3D motion trackers, data-gloves, and joysticks with and without force feedback. The engine also enables the developer to choose the way the scene should be rendered to, i.e. using standard display devices, stereoscopy, or even several simultaneous projection for spatially immersive devices. As part of the evaluation process, we have compared the performance of EnCIMA to a game engine and two scene graph toolkits, through the use of a testbed application. The performance results and the wide variety of non-conventional interaction devices supported are evidences that EnCIMA can be considered a real time virtual reality engine.

Massively multi-user simulations aim to support a large number of users while keeping the communication among the parties synchronous and highly interactive. In this paper, we present a collaborative virtual architecture that supports a large number of users by dividing the virtual environment into multiple adjacent hexagonal regions in order to manage the interest of the entities. A master node, called a hybrid node, constructs a Peer-to-Peer (P2P) overlay network to connect and manage nodes that lie in its region. Messaging is done at the application layer rather than the network layer, and a node-joining algorithm is proposed to reflect the underlying network physical topology onto the data distribution pathways among the end hosts to enhance the system performance. In addition, the introduction of a buffer zone between adjacent zones reduces the number of connections and disconnections that occur when a node frequently moves at the boundary of the two zones and provides more resilience to the system. We also attempt to shift the messaging among parties in one region from a zone-based method to a visibility-driven method to refine their interest by enabling message filtering. The effectiveness of this collaboration architecture is tested through a prototype implementation and a high level application.

In this paper a novel contour tracking algorithm based on a Markovian jump extension of the Kalman filter is presented, we call it the Markovian jump tracker (MJT). Markovian jump linear systems (MJLS) are suitable to model physical systems that behaves linearly but suffers abrupt changes in the dynamics from time to time (accordingly to a Markov chain process) with applications ranging from aircraft control systems to macroeconomics simulation. The tracking method we devised uses Markovian jump filter technology to profit from available multiple linear models -to achieve overall tracker stability-each roughly describing a specific type of expected motion, for example: rotation, translation, cyclic motion, and so on. We verify the method's efficacy in preliminary experiments and compare aspects of its performance with other popular contour tracking algorithm.