Medical and rehabilitation exoskeletons are being increasingly considered by therapists when choosing a treatment for individuals affected by lower limb impairments. Although all such exoskeletons seem to provide similar features and performance, there are, in practice, significant differences among them in terms of maximum walking speed, maximum torque, weight, autonomy, interaction with the user, or even the way to use it. In this review, the state of the art of the main commercial exoskeletons is described, while analyzing their properties, advantages, and disadvantages. Three groups are considered: complete exoskeletons, partial exoskeletons and open lines of research. A comparative analysis between them is performed while considering the main scientific and technical aspects to be improved. In conclusion to this analysis, the balance between feasibility and innovation in exoskeletons development is a design challenge. Commercial exoskeletons must fulfil standards whilst ensuring their safety and robustness. However, achieving a new generation of exoskeletons means a need to implement new hardware paradigms, and to enhance control strategies focused on assist-as-needed scheme. Finally, some aspects to improve current designs of the exoskeleton are presented.

Current exoskeleton designs have opened new ways to therapists and patients for muscle maintenance and strength recovery therapy approaches. However, some issues remain to be solved. Lack of adaptation, versatility, or overweight are some of the main factors that should be improved. In this article, we present a novel design for a functional modular exoskeleton composed of four independent active joints using a decentralized electronic architecture, being the first wearable exoskeleton to apply this technology. This new approach avoids the use of a main processor, and the modular scheme allows every possible configuration between the hip and knee. The decentralized electronic architecture is enabled by implementing adaptive central pattern generator (CPG) algorithms in each module controller. Each CPG has been trained to mimic the joint trajectory of a healthy person walking. The communication between modules assures their coordination through synchronization terms. The operation of each configuration was first tested on a bench at different speeds and later used by a healthy subject. These tests show how the decentralized control strategy maintains the synchronization between modules, reproducing and adapting the trained walking pattern. These results prove that our proposal is suitable for the development of exoskeletons focused on personalized therapies.

Background and purpose: Stroke is one of the leading causes of disability in adults worldwide, and one of the main objectives in the rehabilitation of these patients is to recover the gait. New technologies have emerged to cope with this issue, complementing conventional therapy with the use of devices such as exoskeletons. The Marsi Active Knee (MAK) exoskeleton (Marsi Bionics SL, Madrid, Spain) has already been tested, but an updated version was improved to allow the patients to perform functional exercises. The aim of this study was to assess the safety and usability of the MAK in the stroke population as well as its potential clinical effects.Methods: A single‐group open label intervention trial was conducted. The device was used twice a week for 5 weeks during 1 h per visit. During the visits, sit‐to‐stand transitions, walking, stair climbing, trunk rotations, and weight‐transfer exercises were performed using the device. Adverse events were collected from participants and therapists to assess safety. The Quebec User Evaluation of the Satisfaction with assistive Technology (QUEST 2.0) was used by both therapists and participants to assess usability. To evaluate its clinical effects, active range of motion (ROM) and muscle strength were assessed in the lower limb.Results: Six participants with stroke were recruited. The device was shown to be safe since no serious adverse events were reported neither by patients nor by therapists. Every proposed exercise was performed. Regarding clinical effects, overall muscle strength showed an increase after the treatment, although ROM measurements did not show any difference.Discussion: Our results suggest that the MAK device is safe for stroke patients. Nevertheless, further changes to enhance usability are recommended, such as an improvement of the attachment system and an adaptation for the drop foot. Beneficial effects regarding increases in muscle strength were obtained. Further trials with a larger sample size, longer intervention periods, and a control group are needed to verify these results. Also, future research should focus on the usability of the MAK as an assistive technology.

The integration of Artificial Intelligence (AI) into software testing has emerged as a promising approach to addressing the limitations of conventional methods, particularly in safety-critical domains such as automotive and aerospace. Traditional software testing is often manual, resource-intensive, and prone to incomplete coverage. In contrast, AI-based techniques have been applied to automate test case generation, optimise testing processes, and improve defect prediction. This study extends these approaches to the context of Embodied AI, where the reliability of systems that perceive and act within complex environments is a primary concern. Autonomous satellites are considered as representative embodied agents, for which the robustness of the perception–action loop is mission-critical. The paper presents two main contributions. First, it reviews the state of the art in the application of AI techniques to software testing. Second, it demonstrates the practical viability of these techniques through a proof of concept (PoC) in the satellite domain. In this PoC, generative AI is used to create synthetic datasets simulating adverse atmospheric conditions, including varying levels of cloud opacity. These datasets are used to evaluate and improve an on-board object detection model. Experimental results indicate that this approach reveals limitations of the baseline model and, following retraining, enhances robustness under challenging conditions while maintaining performance in clear scenarios. The findings demonstrate the potential of AI-based testing methodologies to support the validation and improvement of mission-critical embodied systems.

This work presents a comparative evaluation of advanced YOLO architectures for object detection, with a specific focus on their performance in traffic light detection for autonomous driving applications. Two deployment strategies were analyzed: a native implementation using PyTorch and a Model Based Engineering (MBE) implementation through automatic code generation. Evaluation metrics included precision-recall curves and confusion matrices across varying Intersection over Union (IoU) thresholds, as well as mean Average Precision (mAP) to assess detection quality and inference time measurements to evaluate computational efficiency on embedded platforms. The evaluation was based on a custom video extracted from the CARLA simulator, which was meticulously annotated by reviewing each frame to ensure the accuracy of the labeling. The study highlights the compromises between model accuracy and computational cost, providing a reproducible framework for performance benchmarking of object detection algorithms in safety-critical environments.