The safe and continuous operation of Unmanned Surface Vessels (USV) depends on robust visual perception systems capable of detecting and avoiding obstacles in a dynamic and unpredictable maritime environment. The main bottleneck for the development of such systems is the scarcity of comprehensive training datasets that capture the wide range of operational scenarios. This work proposes a programmatic methodology for generating large-scale synthetic datasets using a generative AI-based data augmentation approach. Our pipeline transforms a limited set of real images into a diverse dataset by introducing controlled variations in sea states, weather conditions, lighting, and a wide range of obstacles. The methodology includes modules to ensure image realism, validate the preservation of objects of interest, and mine "hard examples" to focus training on highly complex scenarios. We also propose an experimental design to validate the effectiveness of our generated dataset, comparing the performance of cutting-edge object detection architectures trained with raw data, data augmented by traditional methods, and data generated by our pipeline. The expected results indicate a significant improvement in the accuracy and robustness of the models trained with our method, paving the way for safer and more reliable anti-collision systems on autonomous robotic offshore platforms.

This paper analyzes the integration of sustainable educational robotics and unplugged computing activities as playful practices in pedagogical work with children of the two most elementary age groups in basic education. The goal is to demonstrate the scientific contribution on the relevance of such practices to stimulate cognitive, social, and motor development, based on the democratization of access to technologies and digital education policies, highlighting unplugged computing as a viable strategy for contexts with infrastructure limitations. We use as a case study an implementation project whose objective is to incorporate unplugged computing practices and educational robotics in Early Childhood Education, focusing on children aged 1 year and 7 months to 5 years and 11 months. The actions involved twenty theoretical-practical workshops, with low-cost, playful activities for the development of computational thinking and cognitive skills aligned with the National Common Curriculum Core Base and its Complement in Computing in Basic Education. Practical actions include planning, implementation, and formative evaluation, using workshops, as well as the creation of a public repository of teaching resources. The results show positive impacts on children’s logical reasoning, motor coordination, pattern recognition, problem-solving, and social interaction, in addition to strong family involvement.

The literature on the development of autonomous sailboats lacks reporting the challenges experimented in experiments in real-world settings such as the risk of failure, the complexity of missions, and the absence of established operational procedures. Simulators also lack real-world data, which limits their fidelity. Towards alleviating these problems, this work adopts an exploratory and applied approach based on a longitudinal case study. The operational challenges faced during five years of development and real missions with a robotic sailboat are introduced, proposing operational procedures to reduce risks, increase the agility of field missions, and enable the collection of navigation data. The main contribution is a set of innovative operational practices in missions with robotic sailboats that enables the transition of algorithms from simulation to practical application. An analysis was conducted using data from these practices on real missions, complemented by an evaluation with questionnaires. These new operational procedures increased the safety and effectiveness of field missions.

Current Unmanned Surface Vessel (USV) systems follow the USV-1.0 paradigm (traditional modular perception-planning-control stacks) or more recently the USV-2.0 approach (end-to-end neural policies requiring large training datasets). In contrast to being effective in constrained settings, both paradigms fail in leveraging accumulated human nautical expertise and struggle with data efficiency and interpretability. In this paper we propose a cognitive maritime navigation paradigm that integrates Vision-Language Models (VLMs), Human-in-the-Loop (HITL) learning, and Spatio-Temporal Navigation Memory (STNM) to enable autonomous surface vessels to learn and reuse visuomotor behavior from human demonstrations, which we nominate as USV-3.0. This novel approach integrates three components into a unified cognitive framework: (1) language-conditioned maritime vision-language models (VLMs) that interpret 360 degrees visual scenes and ground perception in the intent conveyed by natural-language mission commands; (2) a confidence-based HITL mechanism that triggers operator intervention when system confidence drops below 90%, storing validated human decisions together with their linguistic and visual context; and (3) an STNM system that records trajectories, visual embeddings, and textual reasoning, enabling hybrid GPS-and CLIP-based retrieval of past behaviors conditioned on mission semantics. We validate USV-3.0 in VisualSimBoat using four real-world navigation scenarios that represent realistic conditions. Considering trained configurations (normal visibility), the system achieves 100% mission success, 98.2% autonomous functioning, and memory retrieval precision within 8m, learning new routes from 23 to 61 HITL episodes-orders of magnitude fewer than typically required by USV-2.0 approaches. Zero-shot transfer experiments under adverse visibility conditions reveal limitations in decision execution, motivating further investigation into safety mechanisms for unseen environments. These results position language-conditioned cognitive navigation as a data-efficient alternative to existing USV paradigms, while clearly delineating its current boundaries and future research directions.

We propose a multi-variate representation approach built with autoencoders, named MVELSA, for underwater image classification. Basically, the idea is that the system instantiates multiple independent autoencoders rather than relying on a single to feature extraction, which allows our system to capture different statistical nuances of the data in the latent space of each autoencoder. The evaluation is performed on the AQUA20 benchmark, which reflects realistic underwater conditions and highly asymmetric class distributions, in which we conducted a comparative study with consolidated deep learning architectures, such as ResNet-18 and YOLOv8n. Experimental results indicate that the convolutional models trained via transfer learning achieve competitive global accuracy but exhibit limited performance on minority classes, as reflected by lower macro-averaged recall and F1-score. In contrast, the MVELSA approach demonstrates balanced performance across classes, under several experimental conditions. Although performance gains are data-dependent, our findings suggest that combining different latent space representations is a promising alternative for underwater image classification, particularly in scenarios with scarce and imbalanced data.

Overall, a big challenge in building a sailboat USV relies on the development of an autonomous system for guidance, navigation, and control (GNC) because both sail and rudder angle must be cooperatively adjusted to correct the navigation direction — traditional propelled boats can be more easily controlled with a straightforward control task to set the rudder angle. Moreover, sailing upwind requires special maneuvers to reach a given target in that unfeasible direction. Reinforcement learning emerges as a promising technique for building autonomous GNCs for sailing robots, but training the neural network with a real sailboat is impractical due to long periods of training and safety reasons. Even traditional control-based approaches are mainly tested in simulated environments due to the difficulties in building and operating a real sailboat. The issue that arises is the fidelity of these simulated environments. In this context, we propose Yara, an oceanic virtual environment with a reliable physics simulation for developing, training, and evaluating autonomous agents to operate digital twins of sailing robots in reinforcement learning and other paradigms. An autonomous sailing robot digital twin is available within the virtual environment, with the foil dynamics constructed based on a real sailing robot. We coupled these foil dynamics in Gazebo’s physics engine to compute the lift and drag forces acting on the sail, rudder, and keel. The simulated world feeds sensors such as cameras, wind sensors, and GPS. The Robot Operating System communicates these sensors’ data through topics, facilitating users’ implementation and testing of new GNC solutions. Yara provides a reliable solution for foil dynamic simulated physics that achieves a simulation speedup of 300 times on an i7 laptop with 8 GB of RAM, powered by a Nvidia RTX 3060 and running Ubuntu 20.04. With this speedup, it is possible to complete a million time steps of deep reinforcement learning training in approximately eight hours. Evaluation scenarios were presented to highlight specific features of the simulator, like the maneuverability of the sailing robot digital twin and applications to train, evaluate, and compare reinforcement learning agents and other control solutions.

Educational Robotics (ER) has emerged as a dy-namic tool to enhance learning in STEAM (Science, Technology, Engineering, Arts, and Mathematics) education, fostering com-petencies, creativity, and problem-solving skills. Latin America, particularly Brazil and Peru, has seen a growing interest in incorporating ER into school curricula. However, the scale and form of implementation differ significantly between these two countries, especially when looking at public and private educational institutions. This article presents a joint analysis of ER policies, practices, and results in public and private schools in Brazil and Peru. Through examining national initiatives, pedagogical methodologies, and teacher training programs, we elucidate the advancements and challenges in integrating ER into educational systems of both countries. We highlight the challenges, innovations, and educational impacts observed in both nations and provide recommendations for equitable and effective expansion.

Learning difficulties in children can have a variety of causes, often interconnected. Factors such as emotional problems (anxiety, stress), communication difficulties, inadequate teaching methodologies, health problems, lack of family support, and even genetic and prenatal factors can contribute to these difficulties. There are several ways to tackle the problem, mainly the use of active methodologies involving technology tools is one of them. In this direction, to help alleviate these problems, this article analyzes the possible use of educational robotics as a support tool to help children with learning difficulties in the teaching-learning process. We seek to answer our main research question, which is to identify if and how educational robotics can help the development of these students' skills and abilities. To do that, we first conducted a systematic review of the literature on the subject, from which we performed an in-depth analysis, including several issues, not just those related to learning difficulties. From this initial stage, we found that there are few studies that address this topic and several open questions. The main gap observed is that many students with learning difficulties are present in schools, but most of them do not participate in robotic activities. Also, in some cases, they show no interest, as will be discussed, on the basis of strictly related works. Given the novelty of the subject and the existence of gaps, we conducted an investigation using a qualitative-quantitative research methodology, using a questionnaire as a data collection tool. The analysis of the results indicates an increase in the use of robotics in education for students with learning difficulties and, above all, that these students benefit from its use. Thus, this study contributes with a step forward in demonstrating that Educational Robotics can help students with learning difficulties.

Educational robotics has been shown to be an effective pedagogical tool for developing various skills and competencies in students, such as computational thinking, problem solving, creativity and teamwork. In this context, teachers are increasingly looking for courses in educational robotics to contribute to the process and development of students. Repo-Educ, a web repository, was developed with the aim of making it easier for teachers and robotics enthusiasts to access and share educational robotics lesson plans, training and encouraging teachers to use robotics effectively in the classroom. In order to expand Repo-Educ’s functionalities and disseminate scientific, technological and intellectual production in educational robotics, this article presents the Intellectual Collection, a new module for Repo-Educ that centralizes and makes available articles, theses, dissertations, software and hardware, kits and other resources developed for the entire educational robotics community and the general public.

We propose a navigation approach for sailboat robots inspired by the decision-making and control processes of experienced human sailors. Our proposed system receives data from a sensor array onboard the sailboat to perceive environmental conditions and make necessary course adjustments. Using these data, the system computes the angular distances for the bow, obstacles, wind, goal, and trajectory to calculate the steering angle. The sail positioning is determined on the basis of the apparent wind angle, which is typically set to be about 20 degrees off from the apparent wind direction. To verify the proposed approach, we performed experiments on a realistic simulation platform, a digital twin named Yara, developed in a previous work. The results demonstrate an improvement of the proposed approach over traditional approaches such as A* and artificial potential field (APF), which are state-of-the-art approaches for trajectory planning and execution in autonomous sailboat navigation.

Minimization of energy consumption with the longest flight time and distance is one of the main challenges in the development of unmanned aerial vehicles (UAV). Vertical take-off and landing (VTOL) is another feature that makes handling such platforms easier. Considering these requirements, we propose the architectural design of a double-hybrid UAV, conceptualized and built as a flying wing with tractor configuration with a combustion engine propeller and an electric multirotor with 3 propeller sets. We performed several tests to validate the proposed UAV model, including take-off, horizontal flight, transition tests, and also tests on autonomy (flight time and distance). As a verification of the tool usefulness, the architectural sketch of our proposal has been granted a patent from the Intellectual Property Institute (Brazil) under the number BR-10-2015-032525-8. The main contribution is the architectural design and conceptualization of the double-hybrid UAV, which consists of easy-to-follow procedural steps that can be reproduced in a simple fashion anywhere.

We introduce the possibility of using a low-cost solar tracker prototype in the context of educational robotics to increase environmental awareness in elementary school students (K12). The developed project uses the Giro-Bot hardware, developed with Arduino coupled with light-sensitive sensors and a servomotor to dynamically track the direction of moving light sources. This simulates the operation of photovoltaic panels with a solar tracking system. More than a technical proposal, this initiative aims to promote interdisciplinary learning in STEAM. It has been developed by undergraduate students as a final project on a robotics discipline in Computer Engineering and Bachelor of Science and Technology courses. The solar tracker has been redesigned as an educational tool that integrates sustainability, maker culture, and educational robotics, with practical experiments that demonstrate pedagogical utility. The results of its applications show the potential to broaden student interest and understanding of topics such as automation, electronics, and environmental awareness through practical experimentation and participant reports.

We propose a methodology for rapid manufacturing of hybrid UAVs considering both very low-cost and easily acquired materials, with rapid prototyping through 3D printing. We adopt a hybrid configuration with fixed-wing and multirotor with the aim of demonstrating that the process is suitable for all categories of UAVs. A fully printed UAV still demands a long printing time, high cost depending on the printing material, and high density, resulting in heavy structures compared to the manufacturing using composite materials. Thus, it is necessary to reconcile low-cost materials available on the market with rapid prototyping technologies such as 3D printing. Our approach can help speed up the process, as the aircraft often has a high acquisition or manufacturing cost in addition to requiring a long manufacturing time. This happens in flight tests, for example, which is the most critical step and prone to accidents that could make the use of the aircraft unfeasible.

This paper presents the latest developmental im-provements done on the URA low-cost robotic kit, which uses a open-source and low-cost robotics paradigm and was designed for general activities in educational robotics. Our kit is planned with the aim of facilitating the teaching of the basic concepts of programming, electronics, and mechanics. The current version (9.0) was redesigned with a focus on accessibility and intuitiveness in order to meet the demands of practical training in the educational context with limited infrastructure. The validation of the developments occurred in a mini-robotics competition held at the School of Science and Technology of the Federal University of Rio Grande do Norte, where freshmen undergraduate students assembled a robot with the kit and faced challenges inspired by the Brazilian Robotics Olympiad (OBR). The results pointed to the effectiveness of the kit in forming student teams, as well as its potential for replication in public schools and robotic events. This paper contributes with the development stages, the components, a simulation software, and programming methodology, pointing out the main difficulties encountered during implementation.

The demand for autonomous Unmanned Surface Vehicles (USVs) is growing across maritime industries, yet there remains a lack of accessible platforms for generating high-quality synthetic datasets needed for AI model development. Real-world data collection is costly, time-consuming, and often impractical in dynamic maritime environments. This work introduces VisualSimBoat, an open-source, high-fidelity simulator built on Unreal Engine that addresses this gap. The simulator offers physics-based water interaction, multi-sensor integration, and tools for creating extensive datasets with synchronized camera imagery, GPS data, and control commands. Experimental results validate its effectiveness in generating data frames for diverse maritime conditions and facilitating streamlined dataset creation. The simulator has proven instrumental in debugging waypoints and leveraging datasets from camera and GPS sensors to train AI models for autonomous navigation.

This paper presents a systematic literature review on teacher training in educational robotics, with a twofold objective: to map the global research landscape and to provide a focused analysis of the Brazilian context. The review was guided by spe-cific research questions and conducted using the Scopus and Web of Science databases for the global perspective, complemented by the Brazilian Digital Library of Theses and Dissertations (BDTD) for the local analysis. Our findings highlight Brazil's significant international contribution, ranking second in the number of publications in both Scopus (6 out of 28) and Web of Science (4 out of 15). At the national level, our analysis identified 11 recent theses and dissertations completed between 2020 and the present. These results indicate that robust methodologies for teacher training in Educational Robotics are not only being actively researched but also effectively implemented throughout Brazil.

Latin American countries have made several efforts to bring Educational Robotics to the education levels of K–12, fostering larger initiatives from governments and private sectors to improve teaching methods. There are important contrasts in the approaches and the effective numbers of implementations, although most countries share specific social-economic challenges. The scale and form differ significantly in such a way that it remains a challenge to analyze and understand the overall implementations in the region. Toward partially answering this question, this article contributes with an analysis of Educational Robotics in Latin American countries based on a systematic review of the scientific research documents on this field present in Scopus and Web of Science, including topics such as methodologies and tools, policies and practices, and main implementations and results. This systematic review aims at first drawing a global panorama and then refining a local (Latin American) panorama on Educational Robotics in K-12. In addition, we provide an status of implementations in main countries by looking at the literature and completed with other documentation such as state documents and formal news. We found that Educational Robotics is being used not only as a technical tool but also as a philosophical, social, and inclusive platform. However, the sector still faces challenges in attracting and retaining talent, aligning curricula with the demands of the 21st century, and ensuring high-quality education. Mostly, funding shortages and disparities in access to the tool remain global concerns in Latin America.

This article analyzes a pedagogical experience carried out with 3- and 4-year-old children from Early Childhood Education, through a pedagogical practice that addresses ocean cleanup, within the scope of the MandacaruBot extension project (EAJ/UFRN). The proposal integrates scientific, environmental, and technological knowledge with the use of sustainable Educational Robotics and unplugged robotics, seeking to promote scientific literacy and computational thinking in a playful and accessible way. The activities involved the construction of artifacts, interactive game, and group discussions, according to the guidelines of the Common National Curriculum Base (BNCC) and the Computing BNCC. In practice, hydraulic arms developed with syringes, gamification, computational thinking, and unplugged robotics were used. The results demonstrated the development of motor, cognitive, and socioemotional skills in children, in addition to engagement with issues of sustainability and citizenship.

This article presents a pedagogical proposal using educational robotics to work with children and adolescents on May 18th - National Day to Combat the Sexual Abuse and Exploitation of Children and Adolescents, strengthening the "Make it Beautiful. Protect our Children" campaign. The proposal is the conclusion of an extension course in educational robotics promoted by the Federal University of Rio Grande do Norte, through the URA project - One Robot Per Student. The intervention is aligned with the national curriculum for basic education, including what it says about computing. As hardware and software, we used sustainable materials, Arduino-UNO and TinkerCad. After its application, we recognize that the proposal can contribute to pedagogical work with educational robotics in the country’s different education networks.