Search

Accurate posture detection of elderly people is crucial to improve monitoring and provide timely alerts in homes and elderly care facilities. Human posture recognition is experiencing a great leap in performance with the incorporation of deep neural networks (DNNs) such as YOLOv9. Unfortunately, DNNs require large amounts of annotated data for training, which can be addressed by using virtual reality images. This paper investigates how to address the appearance domain that lies between synthetic and natural images. Therefore, four experiments (VIRTUAL–VIRTUAL; HYBRID–VIRTUAL; VIRTUAL–REAL; and HYBRID–REAL) were designed to assess the feasibility of recognising the postures of virtual or real elderly people after training with virtual and real images of elderly people. The results show that YOLOv9 achieves the most outstanding accuracy of 98.41% in detecting and discriminating between standing, sitting, and lying postures after training on a large number of virtual images complemented by a much smaller number of real images when testing on real images.

Virtual reality (VR) simulators enable the evaluation of engineering systems and robotic solutions in safe and realistic environments. To do so, VR simulators must run algorithms in real time to accurately recreate the expected behaviour of real-life processes. This work was aimed at determining a suitable configuration for processing images taken from a virtual unmanned aerial vehicle developed in Unity using OpenCV. To this end, it was focused on comparing two approaches to integrate video processing in order to avoid potential pitfalls such as delays and bottlenecks. The first approach used a dynamic link library (DLL) programmed in C++, and the second an external module programmed in Python. The native DLL ran internally on the same Unity thread, as opposed to the Python module that ran in parallel to the main process and communicated with Unity through the Message Queue Telemetry Transport (MQTT) protocol. Pre-transmission processing, data transmission and video processing were evaluated for a pair of typical image-processing tasks like colour and face detection. The analysis confirmed that running the Python module in parallel does not overload the main Unity thread and achieves better performance than the C++ plugin in real-time simulation.

Some daily tasks involve grasping multiple objects in one hand and releasing them in a determined order, for example laying out a surgical table or distributing items on shelves. For training these tasks in Virtual Reality (VR), there is no technique for allowing users to grasp multiple objects in one hand in a realistic way, and it is not known if such a technique would benefit user experience. Here, we design a multi-object grasp technique that enables users to grasp multiple objects in one hand and release them in a controlled way. We tested an object placement task under three conditions: real life, VR with single-object grasp and VR with multi-object grasp. Task completion time, distance travelled by the hands and subjective experience were measured in three scenarios: sitting in front of a desktop table, standing up in front of shelves and a room-size scenario where walking was required. Results show that the performance in a real environment is better than in Virtual Reality, both for single-object and multi-object grasping. The single-object technique performs better than the multi-object, except for the room scenario, where multi-object leads to less distance travelled and reported physical demand. For use cases where the distances are small (i.e., desktop scenario), single-object grasp is simpler and easier to understand. For larger scenarios, the multi-object grasp technique represents a good option that can be considered by other application designers.

Continued advances in machine learning (ML) and deep learning (DL) present new opportunities for use in a wide range of applications. One prominent application of these technologies is defect detection and classification in the manufacturing industry in order to minimise costs and ensure customer satisfaction. Specifically, this scoping review focuses on inspection operations in the semiconductor manufacturing industry where different ML and DL techniques and configurations have been used for defect detection and classification. Inspection operations have traditionally been carried out by specialised personnel in charge of visually judging the images obtained with a scanning electron microscope (SEM). This scoping review focuses on inspection operations in the semiconductor manufacturing industry where different ML and DL methods have been used to detect and classify defects in SEM images. We also include the performance results of the different techniques and configurations described in the articles found. A thorough comparison of these results will help us to find the best solutions for future research related to the subject.

Holographic methods from optics can be adapted to acoustics for enabling novel applications in particle manipulation or patterning by generating dynamic custom-tailored acoustic fields. Here, we present three contributions towards making the field of acoustic holography more widespread. Firstly, we introduce an iterative algorithm that accurately calculates the amplitudes and phases of an array of ultrasound emitters in order to create a target amplitude field in mid-air. Secondly, we use the algorithm to analyse the impact of spatial, amplitude and phase emission resolution on the resulting acoustic field, thus providing engineering insights towards array design. For example, we show an onset of diminishing returns for smaller than a quarter-wavelength sized emitters and a phase and amplitude resolution of eight and four divisions per period, respectively. Lastly, we present a hardware platform for the generation of acoustic holograms. The array is integrated in a single board composed of 256 emitters operating at 40 kHz. We hope that the results and procedures described within this paper enable researchers to build their own ultrasonic arrays and explore novel applications of ultrasonic holograms.

Personal assistant robots provide novel technological solutions in order to monitor people’s activities, helping them in their daily lives. In this sense, unmanned aerial vehicles (UAVs) can also bring forward a present and future model of assistant robots. To develop aerial assistants, it is necessary to address the issue of autonomous navigation based on visual cues. Indeed, navigating autonomously is still a challenge in which computer vision technologies tend to play an outstanding role. Thus, the design of vision systems and algorithms for autonomous UAV navigation and flight control has become a prominent research field in the last few years. In this paper, a systematic mapping study is carried out in order to obtain a general view of this subject. The study provides an extensive analysis of papers that address computer vision as regards the following autonomous UAV vision-based tasks: (1) navigation, (2) control, (3) tracking or guidance, and (4) sense-and-avoid. The works considered in the mapping study—a total of 144 papers from an initial set of 2081—have been classified under the four categories above. Moreover, type of UAV, features of the vision systems employed and validation procedures are also analyzed. The results obtained make it possible to draw conclusions about the research focuses, which UAV platforms are mostly used in each category, which vision systems are most frequently employed, and which types of tests are usually performed to validate the proposed solutions. The results of this systematic mapping study demonstrate the scientific community’s growing interest in the development of vision-based solutions for autonomous UAVs. Moreover, they will make it possible to study the feasibility and characteristics of future UAVs taking the role of personal assistants.