With the increase in elderly population throughout the world, especially in the European Union, there is an utmost need for assistive technologies to help them achieving daily tasks in an independent or collaborative manner. This paper presents a novel sensorless control for an active assistive walker featuring a disturbance observer for the estimation of the force applied by the user (thus eliminating the need of external force sensors) and admittance control for smooth locomotion. The experimental validation is carried out in the ROS2 framework for data computation and analysis. The walker, driven by motorized wheels, is tested on flat and inclined surfaces. The results demonstrate accurate force estimation, efficient motion control, and stable assistance with a balanced tuning of the admittance control parameters, i.e. virtual mass and damping, thus ensuring responsive and smooth behavior of the system.

Autonomous navigation in unstructured outdoor environments presents significant challenges due to the complexity and variability of terrain and obstacles. Traditional navigation methods struggle with adaptability, while learning-based approaches, particularly deep Reinforcement Learning (RL), have shown promise but faces difficulties in generalization and data efficiency. To address these challenges, we tested the performance of various deep RL algorithms for point-goal navigation, on MIDGARD, our photorealistic simulation platform built on Unreal Engine. We compare PPO, A2C, SAC, and TD3, evaluating their effectiveness based on success rates and reward progression. Our results indicate that PPO outperforms other algorithms, achieving the highest success rate, while off-policy methods struggle due to inefficient exploration and policy updates.

Being able to estimate the traversability of the area surrounding a mobile robot is a fundamental task in the design of a navigation algorithm. However, the task is often complex, since it requires evaluating distances from obstacles, type and slope of terrain, and dealing with non-obvious discontinuities in detected distances due to perspective. In this paper, we present an approach based on deep learning to estimate and anticipate the traversing score of different routes in the field of view of an on-board RGB camera. The backbone of the proposed model is based on a state-of-the-art deep segmentation model, which is fine-tuned on the task of predicting route traversability. We then enhance the model's capabilities by a) addressing domain shifts through gradient-reversal unsupervised adaptation, and b) accounting for the specific safety requirements of a mobile robot, by encouraging the model to err on the safe side, i.e., penalizing errors that would cause collisions with obstacles more than those that would cause the robot to stop in advance. Experimental results show that our approach is able to satisfactorily identify traversable areas and to generalize to unseen locations.

NA

Communication networks for cooperative mobile robot applications have to cope with some peculiar requirements that make them quite different from typical MANETs or WSNs. Recent cooperative robot applications envisage the support of low datarate communication technologies, as this choice is beneficial in terms of energy consumption, weight reduction and integration with WSNs. In this context, this paper discusses the design choices behind a novel protocol for cooperating mobile robot applications that operates on low datarate networks and is able to provide bounded latencies, mobility and reliability.

This paper proposes an augmented reality visualization interface to simultaneously present visual and laser sensors information further enhanced by stereoscopic viewing and 3-D graphics. The use of graphic elements is proposed to represent laser measurements that are aligned to video information in 3-D space. This methodology enables an operator to intuitively comprehend scene layout and proximity information and so to respond in an accurate and timely manner. The use of graphic elements to assist teleoperation, sometime discussed in the literature, is here proposed following an innovative approach that aligns virtual and real objects in 3-D space and color them suitably to facilitate comprehension of objects proximity during navigation. This paper is developed based on authors' previous experience on stereoscopic teleoperation. The approach is experimented on a real telerobotic system, where a user operates a mobile robot located several kilometers apart. The result showed simplicity and effectiveness of the proposed approach.

The use of 3-D stereoscopic visualization may provide a user with higher comprehension of remote environments in teleoperation when compared with 2-D viewing, in particular, a higher perception of environment depth characteristics, spatial localization, remote ambient layout, faster system learning, and decision performance. Works in the paper have demonstrated how stereo vision contributes to the improvement of the perception of some depth cues, often for abstract tasks, while it is hard to find works addressing stereoscopic visualization in mobile robot teleguide applications. This paper intends to contribute to this aspect by investigating the stereoscopic robot teleguide under different conditions, including typical navigation scenarios and the use of synthetic and real images. This paper also investigates how user performance may vary when employing different display technologies. Results from a set of test trials run on seven virtual reality systems, from laptop to large panorama and from head-mounted display to Cave automatic virtual environment (CAVE), emphasized few aspects that represent a base for further investigations as well as a guide when designing specific systems for telepresence.

The aim of this paper is to investigate on the closed loop welding controller of a rapid manufacturing shaped metal deposition (SMD) process. SMD was developed and patented by Rolls-Royce in order to produce mechanical parts directly from a CAD model. The paper describes a deep investigation upon the welding dynamics and parameters in order to develop an SMD automatic controller. An SMD plant has been set up as described in the paper, and some experimental work pieces have been produced. On the basis of the experiences acquired, some basic control laws have been carried out, and a closed loop controller has been implemented, using image processing and a video feedback control. The experimental results reported confirm the validity of the proposed strategy.

This paper describes the architecture of a UAV designed to study the composition of gas inside volcanic plumes. The main aim of the system is that of flying inside the plume (volcanic cloud) to directly analyze the concentration of the main components of the fumes. The system must be capable of flying autonomously at up to 4000m altitude with a payload of 5Kg using electric motors, to avoid contaminations with the gas sampling system, at a cruise speed of 40km/h. The complete architecture is presented together with the HMI: this can be used both for path planning and for navigation.

In this paper a traction control algorithm for mobile robots in outdoor environment is presented. The algorithm using a wheel slip estimator optimizes the traction in sandy terrains like those found near some volcanic vents. Some comparative tests with a PI control using a simulator show the adherence improvements. In order to analyze the algorithm efficiency in real conditions, several comparative tests have been executed on the P6W robot: a small robot with six independently actuated wheels.

This work describes an algorithm for the localization and self calibration of a mobile robot. Data from a DGPS and two optical encoders are fused by an extended Kalman filter (EKF) in order to calculate the absolute position of the robot and to estimate the values of the parameters used by the odometry (wheels radii and wheelbase). The presented self calibrating EKF (EKF/sub SC/) has been tested both on a simulator and on the mobile robot "Robovolc": a six independently actuated wheels robot for volcano exploration. The EKF/sub SC/ does not need a separate phase for the parameter identification. Moreover the odometry parameters of the Robovolc robot have been identified both with the UMBmark and with the EKF/sub SC/; some comparative tests show an improvement of the estimated trajectories.

A control strategy for a two wheels planar robot that distributes the motor torques to optimize the traction is presented. A comparison between this strategy and PI control is described using the tests results obtained through a dynamic simulator. This work also shows how the proposed control strategy improves the traction in most situations with respect to a simple PI control, but when the adherence is lost the PI control is more reliable. Finally, this work examines the problems of this strategy and suggests how to solve them.

This paper presents a dynamics analysis of a mobile robot, named M6, developed for volcanic environment exploration. The architecture of this kind of robot has six wheels which are coupled to the chassis by means of revolute joints. The robot is built by the DEES of Catania University. A multibody model of the system was constructed in a calculation code. Particular attention was given to the choice of tyre model to simulate the real tyre-road contact. A comparison of the simulation and experimental results confirmed the good agreement between the model and the real system.

In this paper the problem of controlling the attitude of a rigid body, such as a Spacecraft, in three-dimensional space is approached by introducing two new control strategies developed in hypercomplex algebra. The proposed approaches are based on two parallel controllers, both derived in quaternion algebra. The first is a feedback controller of the proportional derivative (PD) type, while the second is a feedforward controller, which is implemented either by means of a hypercomplex multilayer perceptron (HMLP) neural network or by means of a hypercomplex radial basis function (HRBF) neural network. Several simulations show the performance of the two approaches. The results are also compared with a classical PD controller and with an adaptive controller, showing the improvements obtained by using neural networks, especially when an external disturbance acts on the rigid body. In particular the HMLP network gave better results when considering trajectories not presented during the learning phase.

In this work a neural network for the prediction of the complex and nonlinear behaviour of a catalytic reforming of a refinery has been developed. In a fuel refinery reforming is a conversion process to increase the octane number of the desulphurated heavy naphtha in charge. The neural model has been trained and validated on experimental measurements. The results confirmed the suitability of the proposed approach.

Natural circulation loops represent important elements of many technologically relevant systems. For this reason, their instability represents a major concern, as it consists of oscillations leading to flow reversal. The analysis of these processes was addressed in various theoretical works, mainly based on mathematical approaches to the problem. The models obtained in this way suffers a poor correspondence between simulated and experimental data. To solve this problem, the identification of the system was adopted in this paper, and a neural network model was obtained by means of input-output measurements detected on an experimental natural circulation loop. Moreover, the neural model was used in a predictive scheme, in order to allow long term prediction of the birth of unstable behaviors.

A new type of micro walking robot named PLIF (piezo light intelligent flea) is introduced. These robots, that walk by using piezoceramic legs, are very small in size, but, at the same time, fast and agile. Three different types of PLIF have been designed and built and several dynamic measures have been performed. Moreover a self-learning technique has been implemented and tested in order to increase the autonomy of these systems.

In the paper a strategy in order to decrease the learning time without affecting the learning efficiency for a complex valued Multi Layer Perceptron (CMLP) is proposed. The methodology makes use of auxiliary devices that enable one to predict the connections trend of the principal neural network whose learning phase is not damaged since the auxiliary devices run concurrently with it. A numerical example is reported which shows the suitability of the proposed approach.