Abstract This article describes the methodology used to train and test a Deep Neural Network (DNN) with Photoplethysmography (PPG) data performing a regression task to estimate the Respiratory Rate (RR). The DNN architecture is based on a model used to infer the heart rate (HR) from noisy PPG signals, which is optimized to the RR problem using genetic optimization. Two open-access datasets were used in the tests, the BIDMC and the CapnoBase. With the CapnoBase dataset, the DNN achieved a median error of 1.16 breaths/min, which is comparable with analytical methods in the literature, in which the best error found is 1.1 breaths/min (excluding the 8 % noisiest data). The BIDMC dataset seems to be more challenging, as the minimum median error of the literature’s methods is 2.3 breaths/min (excluding 6 % of the noisiest data), and the DNN based approach achieved a median error of 1.52 breaths/min with the whole dataset.

Resumen Este trabajo presenta una metodología para entrenar y probar una red neuronal profunda (Deep Neural Network – DNN) con datos de fotopletismografías (Photoplethysmography – PPG), con la finalidad de llevar a cabo una tarea de regresión para estimar la frecuencia respiratoria (Respiratory Rate – RR). La arquitectura de la DNN está basada en un modelo utilizado para inferir la frecuencia cardíaca (FC) a partir de señales PPG ruidosas. Dicho modelo se ha optimizado a través de algoritmos genéticos. En las pruebas realizadas se usaron BIDMC y CapnoBase, dos conjuntos de datos de acceso abierto. Con CapnoBase, la DNN logró un error de la mediana de 1,16 respiraciones/min, que es comparable con los métodos analíticos reportados en la literatura, donde el mejor error es 1,1 respiraciones/min (excluyendo el 8 % de datos más ruidosos). Por otro lado, el conjunto de datos BIDMC aparenta ser más desafiante, ya que el error mínimo de la mediana de los métodos reportados en la literatura es de 2,3 respiraciones/min (excluyendo el 6 % de datos más ruidosos). Para este conjunto de datos la DNN logra un error de mediana de 1,52 respiraciones / min

Abstract Introduction Autism Spectrum Disorder is a set of developmental disorders that imply in poor social skills, lack of interest in activities and interaction with people. Treatments rely on teaching social skills and in such therapies robotics may offer aid. This work is a pilot study, which aims to show the development and usage of a ludic mobile robot for stimulating social skills in ASD children. Methods A mobile robot with a special costume and a monitor to display multimedia contents was designed to interact with ASD children. A mediator controls the robot’s movements in a room prepared for interactive sessions. Sessions are recorded to assess the following social skills: eye gazing, touching the robot and imitating the mediator. The interaction is evaluated using the Goal Attainment Scale and Likert scale. Ten children were evaluated (50% with ASD), using as inclusion criteria children with age 7-8, without use of medication, and without tendency to aggression or stereotyped movements. Results It was observed that the ASD group touched the robot about twice more in average than the control group (CG). They also looked away and imitated the mediator in a quite similar way as the CG, and showed extra social skills (verbal and non-verbal communication). These results are considered an advance in terms of improvement of social skills in ASD children. Conclusions Our studies indicate that the robot stimulated social skills in 4/5 of the ASD children, which shows that its concepts are useful to improve socialization and quality of life.

This work discusses the use of optical flow to generate the sensorial information a mobile robot needs to react to the presence of obstacles when navigating in a non-structured environment. A sensing system based on optical flow and time-to-collision calculation is here proposed and experimented, which accomplishes two important paradigms. The first one is that all computations are performed onboard the robot, in spite of the limited computational capability available. The second one is that the algorithms for optical flow and time-to-collision calculations are fast enough to give the mobile robot the capability of reacting to any environmental change in real-time. Results of real experiments in which the sensing system here proposed is used as the only source of sensorial data to guide a mobile robot to avoid obstacles while wandering around are presented, and the analysis of such results allows validating the proposed sensing system.