Journal of Mechatronics Engineering

Comparative analysis of localization methods for autonomous mobile robots using Robot Operating System 2

Antônio Sávio Silva Oliveira, Josias Guimarães Batista — Mon, 24 Nov 2025 00:00:00 +0000

Localization is a fundamental requirement for mobile robots. In order to navigate autonomously and perform tasks, a robot must accurately estimate its position within the environment over time. This work aims to analyze and compare various mobile robot localization methods identified during the literature review, including Monte Carlo Localization (MCL), Adaptive Monte Carlo Localization (AMCL), Sensor Fusion, and a combined method of AMCL with Sensor Fusion, all implemented using the Robot Operating System (ROS 2). The study was carried out in the Webots simulator, with the algorithms developed in Python. Each method was evaluated in terms of efficiency (location accuracy) and performance (hardware resource consumption). The combined method of AMCL with Sensor Fusion achieved the best performance in terms of position accuracy, with a root mean squared error (RMSE) of 4–5 cm and an R² score ranging from 95.10\% to 99.79\% in Scenario 01, and from 68.20\% to 89.65\% in Scenario 02. The Sensor Fusion method ranked second, with an average error of 4–7 cm and an R² score of 94.22\% to 99.69\% in Scenario 01, and 52.27\% to 81.42\% in Scenario 02. Regarding hardware usage, Sensor Fusion showed the lowest resource consumption, using around 17\% of CPU and 32 MB of RAM, followed by AMCL, which used 21\% of CPU and 40 MB of RAM. The main contributions of this work include: the application and evaluation of different localization techniques in specific simulation scenarios, allowing for a comparative study; the use of ROS 2 and the public availability of the developed algorithms and results in a GitHub repository, supporting further studies in Webots simulation, ROS 2, and robot localization techniques.

Comparison of Algorithms for Path Planning Collision Avoidance

Mon, 24 Nov 2025 00:00:00 +0000

Collision-free path planning is a critical aspect of mobile robotic navigation, with significant applications in autonomous systems and intelligent transportation. The choice of an algorithm directly influences computational efficiency and trajectory quality. This study evaluates three widely used paradigms: Artificial Potential Fields (APF), Probabilistic Roadmap Method (PRM), and Rapidly-Exploring Random Tree (RRT). Experiments were conducted in a two-dimensional environment with two and three stationary obstacles, assessing each method based on execution time and path length. The results indicate that APF is simple and fast but prone to local minima. PRM is effective for complex environments but comes with higher computational costs. RRT efficiently explores space, but often generates nonoptical trajectories. The best approach depends on environmental constraints and computational requirements. For future work, increasing scenario complexity or validating the results through real-world robotic experiments is recommended.

Detection of Microcalcifications in Mammography using Image Processing

Mon, 24 Nov 2025 00:00:00 +0000

Breast cancer is the most diagnosed type of cancer and the one that causes the most deaths in women in the world. Mammograms allow the detection of microcalcifications at an early stage. The objective of this work is to develop a computer vision system to detect microcalcifications in mammography images, for this purpose images from the Curated Breast Imaging Subset of Digital Database for Screening Mammography (CBIS-DDSM) database are used. The detection algorithm is divided into two parts, preprocessing and segmentation. A region of interest was not defined, as it is considered that the entire breast is subject to the appearance of microcalcifications. The programming language used is Python, together with the Numpy and OpenCV libraries. For validation, accuracy, sensitivity, specificity, positive predictive value and Dice Similarity were calculated. The experiments show that the accuracy of the method is 98.2\%, the sensitivity is 49.6\%, the specificity is 89.3\%, the positive predictive value is 78.7\% and the Similarity Dice is 52.7\%. The developed system achieved the desired objective, with good performance.

Evaluation of synchronizer geometric and tribological parameters and its influences to phases of synchronization through simulation in GT-Suite.

Rômulo do Nascimento Rodrigues, Gabriela Achtenová, Lukáš Kazda — Mon, 24 Nov 2025 00:00:00 +0000

The transmission system is a vital component that significantly affects a vehicle's performance, power, and fuel efficiency. As vehicle technologies advance, demands on transmission systems continue to increase, with performance often evaluated through gear efficiency, noise levels, and the comfort of gear shifts. Among the key parameters for transmission system performance, synchronization time plays a crucial role. While previous studies have proposed mathematical and computational models of synchronization processes validated by experimental data, limited work has employed GT-Suite software. This article models a gear shift mechanism in GT-Suite to explore the synchronization process phases and analyze parameters influencing synchronization time. The study examines structural and tribological design parameters, evaluating their individual and collective impact on synchronization time. The results demonstrate the influence of varying parameters on the synchronization process, identifying critical phases where each parameter has the most significant effect. These findings provide valuable insights into optimizing synchronization mechanisms and reducing gear-shifting time. Future work may focus on expanding the parameter range and integrating experimental validation to enhance the model's applicability further.