With the advent of the Fourth Industrial Revolution, Internet of Things (IoT) and robotic systems are closely cooperating, reshaping their relations and managing to develop new-generation devices. Such disruptive technology corresponds to the backbone of the so-called Industry 4.0. The integration of robotic agents and IoT leads to the concept of the Internet of Robotic Things, in which innovation in digital systems is drawing new possibilities in both industrial and research fields, covering several domains such as manufacturing, agriculture, health, surveillance, and education, to name but a few. In this manuscript, the state-of-the-art of IoRT applications is outlined, aiming to mark their impact on several research fields, and focusing on the main open challenges of the integration of robotic technologies into smart spaces. IoRT technologies and applications are also discussed to underline their influence in everyday life, inducing the need for more research into remote and automated applications.

In recent years, the number of older people living alone has increased rapidly. Innovative vision systems to remotely assess people’s mobility can help healthy, active, and happy aging. In the related literature, the mobility assessment of older people is not yet widespread in clinical practice. In addition, the poor availability of data typically forces the analyses to binary classification, e.g. normal/anomalous behavior, instead of processing exhaustive medical protocols. In this paper, real videos of elderly people performing three mobility tests of a clinical protocol are automatically categorized, emulating the complex evaluation process of expert physiotherapists. Videos acquired using low-cost cameras are initially processed to obtain skeletal information. A proper data augmentation technique is then used to enlarge the dataset variability. Thus, significant features are extracted to generate a set of inputs in the form of time series. Four deep neural network architectures with feedback connections, even aided by a preliminary convolutional layer, are proposed to label the input features in discrete classes or to estimate a continuous mobility score as the result of a regression task. The best results are achieved by the proposed Conv-BiLSTM classifier, which achieves the best accuracy, ranging between 88.12% and 90%. Further comparisons with shallow learning classifiers still prove the superiority of the deep Conv-BiLSTM classifier in assessing people’s mobility, since deep networks can evaluate the quality of test executions.

Human motion tracking is a valuable task for many medical applications. Marker-based optoelectronic systems are considered the gold standard in human motion tracking. However, their use is not always feasible in clinics and industrial environments. On the other hand, marker-less sensors became valuable tools, as they are inexpensive, noninvasive and easy to use. However, their accuracy can depend on many factors including sensor positioning, light conditions and body occlusions. In this study, following previous works on the feasibility of marker-less systems for human motion monitoring, we investigate the performance of the Microsoft Azure Kinect sensor in computing kinematic and dynamic measurements of static postures and dynamic movements. According to our knowledge, it is the first time that this sensor is compared with a Vicon marker-based system to assess the best camera positioning while observing the upper body part movements of people performing several tasks. Twenty-five healthy volunteers were monitored to evaluate the effects of the several testing conditions, including the Azure Kinect positions, the light conditions, and lower limbs occlusions, on the tracking accuracy of kinematic, dynamic, and motor control parameters. From the statistical analysis of the performed measurements, the camera in the frontal position was the most reliable, the lighting conditions had almost no effects on the tracking accuracy, while the lower limbs occlusion worsened the accuracy of the upper limbs. The assessment of human static postures and dynamic movements based on experimental data proves the feasibility of applying the Azure Kinect to the biomechanical monitoring of human motion in several fields.

Nowadays, the need for reliable and low-cost multi-camera systems is increasing for many potential applications, such as localization and mapping, human activity recognition, hand and gesture analysis, and object detection and localization. However, a precise camera calibration approach is mandatory for enabling further applications that require high precision. This paper analyzes the available two-camera calibration approaches to propose a guideline for calibrating multiple Azure Kinect RGB-D sensors to achieve the best alignment of point clouds in both color and infrared resolutions, and skeletal joints returned by the Microsoft Azure Body Tracking library. Different calibration methodologies using 2D and 3D approaches, all exploiting the functionalities within the Azure Kinect devices, are presented. Experiments demonstrate that the best results are returned by applying 3D calibration procedures, which give an average distance between all couples of corresponding points of point clouds in color or an infrared resolution of 21.426 mm and 9.872 mm for a static experiment and of 20.868 mm and 7.429 mm while framing a dynamic scene. At the same time, the best results in body joint alignment are achieved by three-dimensional procedures on images captured by the infrared sensors, resulting in an average error of 35.410 mm.

Falls represent one of the most serious clinical problems in the elderly population. This risk is even more important in people suffering from neurodegenerative problems. This work aims to instrumentally assess the balance performance of elderly people and specifically those suffering from neurodegenerative diseases, to obtain an objective evaluation of their risk of falls. This paper presents a vision-based system made of three low-cost cameras, able to automatically infer important mobility parameters by observing the execution of well-established tests for stability assessment. This result is achieved by a dedicated image processing pipeline, which processes videos to get dynamic user skeletons, and the following strategy for information management, which targets to feature extraction. This information finally feeds a classifier, namely a decision tree, trained to predict the risk of fall of patients within 5 classes of interest. Actual experiments performed on actual video recordings prove a good agreement of results with those expected, labeled by expert therapists, with final prediction accuracy of 79.1%.