One of the studies of the active brain-computer interface (BCI) focuses on identifying movements from human neurophysiological signals to control external devices such as robotic arms. In the literature, EEG-based BCI is utilized to decode user’s information to perform action or fill the gap from the brain to the arms in the case of illness. The purpose of this work is to understand, among the scientific literature, what the best Deep Learning (DL) architecture is for motor execution (ME) classification. Data from 105 people from the Physionet dataset and 15 subjects from the Upper Limb dataset were used. EEGnetv4, Deep4Net, and EEGITnet were used to classify EEG signals under ME for real-time BCI. The best results were achieved from the EEGNET trained without Common Spatial Pattern transformation, for both datasets.

This paper illustrates some remarkable properties of linear time-varying discrete-time positive systems with delays. First, we discuss how a well known property of positive delay systems with time-invariant matrices, namely the equivalence among delay-dependent and delay-independent stability, does not generalize to positive delay systems when the system matrices are time-varying. Then, we illustrate how a stability analysis based on the existence of linear co-positive Lyapunov functions on the zero-delay system and on its dual produces a remarkable dissimilarity to what happens in the time-invariant case: the dual condition is sufficient to prove delay-independent stability, whereas the primal is not even a stability condition. Implications and further results are discussed.

Providing efficient yet accurate statistical models is a challenging problem in many applications. When elementary models are not sufficiently descriptive, mixtures of densities can be used. A complexity management issue arises when mixture models are employed: the number of components should be a trade-off between the complexity and the accuracy of the model. However, in general, it is not obvious how to determine the right number of mixture components for a specific application. In a previous work, theoretical foundations to address such a topic have been laid, grounded on the use of the Composite Transportation Dissimilarity between mixtures, and a preliminary criterion to manage the complexity of a mixture model has been proposed. In this paper, additional theoretical insights are provided that allow to formulate a novel adaptive mixture reduction algorithm. Numerical tests show that in most cases the new algorithm constitutes a significant improvement over the previous one.

This paper addresses the problem of glucose control in the presence of sampled measurements. This topic is important in the study and development of the Artificial Pancreas (AP), which is a general expression to describe a set of techniques for the control of the glucose behaviour by means of exogenous insulin administration in diabetic individuals. Differently from most of the approaches available in the literature, we not only assume the lack of insulin measurements, but also the availability of the glucose measurements just at sampling times. An observer is designed for the model-based reconstruction of glucose and insulin trajectories from the glucose samples. On top of that, a feedback algorithm (based on the estimated state) is designed to continuously deliver exogenous intra-venous insulin to the patient. Simulations have been performed in-silico on models of virtual patients identified from real data and in the presence of quantization errors. The preliminary results highlight the potential of the proposed approach.

This paper introduces a new analytical algorithm to perform the localization of a mobile robot using odometry and laser readings. Based on a polynomial approach, the proposed algorithm provides the optimal affine filter in a class of suitably defined estimators. The performance of the algorithm has been evaluated through simulations. The comparison with the standard Extended Kalman Filter shows that the proposed filter provides good estimates also in critical situations where the system nonlinearities cause a bad behavior of the EKF.

A control-oriented modeling approach for describing kinematics and dynamics of robots in contact with a dynamic environment is presented. In many robotic tasks the manipulator in contact cannot be simply modeled as a kinematically constrained system. Conversely, modeling of robot-environment interactions through dynamic impedance may not fit the task layout. A suitable model structure is proposed in this note that handles the more general case in which purely kinematic constraints on the robot end-effector live together with dynamic interactions. Feasible end-effector configurations are parameterized from the environment point of view, using a minimal set of coordinates. Accordingly, a description is obtained also for admissible velocities and contact forces. In particular, a force parameterization is chosen so as to separate static reaction forces from active forces responsible for energy transfer between robot and environment. The overall dynamics of the coupled robot-environment system is obtained in a single framework. The introduced modeling technique naturally leads to the design of new hybrid control laws.>

Some issues on the problem of filtering force and velocity measures, in hybrid control, for recovering their consistence with an available environment model are discussed. Since measured forces and velocities in practice do not agree exactly with the model, some filtering procedure is required for recovering consistence, before the control phase. The filtering process is addressed as the resolution of an algebraic overdetermined linear system, solvable approximately by means of weighted pseudoinversion. A basic issue is shown to be the choice of the weight matrix that, if properly defined, allows the solution of the theoretical difficulties. Examples clarify the theory.>