Vicente Arévalo
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Vicente Arévalo has a total of 52 co-authors. Here are the top co-authors by number of shared publications:
- Javier Gonzalez-Jimenez
Universidad de Málaga
15 shared publications - Ana Cruz-Martín
Universidad de Málaga
5 shared publications C. Galindo
Department of System Engineering and Automation, University of Malaga, Spain
4 shared publications- Cipriano Galindo
Universidad de Málaga
4 shared publications - Juan Manuel Gandarias
Universidad de Málaga
4 shared publications - Juan-Antonio Fernández-Madrigal
Universidad de Málaga
4 shared publications
Latest Publications
- Open Access
- 1 September 2026
- Published by Universidade da Coruna in Jornadas de Automática
- No. 47
Una aproximación preliminar al despliegue asíncrono de software de aprendizaje por refuerzo en sistemas físicos
Reinforcement Learning (RL) usually assume fixed timesteps, ignoring temporal discrepancies between simulation and reality and the effects of system latencies. These factors can degrade performance in sim-to-real transfer. This work introduces a fully decoupled solution enabling asynchronous, algorithm-agnostic, and time-aware RL deployment. Our approach isolates the algorithm from the agent into independent loops, overcoming the limitations of traditional synchronous frameworks. This facilitates integration with physical robots and closed-source simulators, enabling variable-time RL strategies. Our tool is compatible with libraries like Stable-Baselines3, simulators such as MuJoCo, and robots like the Franka Emika Panda. A preliminary experimental validation has demonstrated its effectiveness with such robotics manipulator, and an initial version of the software has been made available to the community on GitHub: https://github.com/uncore-team/rl_spin_decoupler.
- 1 September 2026
- Published by Universidade da Coruna in Jornadas de Automática
- No. 47
Open AccessUna aproximación preliminar al despliegue asíncrono de software de aprendizaje por refuerzo en sistemas físicos
Reinforcement Learning (RL) usually assume fixed timesteps, ignoring temporal discrepancies between simulation and reality and the effects of system latencies. These factors can degrade performance in sim-to-real transfer. This work introduces a fully decoupled solution enabling asynchronous, algorithm-agnostic, and time-aware RL deployment. Our approach isolates the algorithm from the agent into independent loops, overcoming the limitations of traditional synchronous frameworks. This facilitates integration with physical robots and closed-source simulators, enabling variable-time RL strategies. Our tool is compatible with libraries like Stable-Baselines3, simulators such as MuJoCo, and robots like the Franka Emika Panda. A preliminary experimental validation has demonstrated its effectiveness with such robotics manipulator, and an initial version of the software has been made available to the community on GitHub: https://github.com/uncore-team/rl_spin_decoupler.
- Open Access
- 1 September 2026
- Published by Universidade da Coruna in Jornadas de Automática
- No. 47
Análisis del tiempo de paso en el aprendizaje por refuerzo para robots móviles
The use of Deep Reinforcement Learning (DRL) in robotics has gained prominence in the last decade, but the impact of its intrinsic temporal components of DRL when applied to physical systems - generally summarized as the timestep duration -, remains underexplored. While some approaches to this issue include timestep duration as part of action space, they offer a limited analysis, overlooking key elements such as software architecture asynchrony or system-induced latencies. These factors cause deviations between nominal and actual timesteps, negatively impacting learning. This article proposes a set of navigation tasks for mobile robots specifically designed to analyze the effects of timestep duration, supported by over 2000 hours of simulation. To our knowledge, this constitutes the first multimetric evaluation of the influence of step time duration on DRL-based robotic tasks. The findings confirm the existence of optimal step times that maximize learning efficiency and task completion.
- 1 September 2026
- Published by Universidade da Coruna in Jornadas de Automática
- No. 47
Open AccessAnálisis del tiempo de paso en el aprendizaje por refuerzo para robots móviles
The use of Deep Reinforcement Learning (DRL) in robotics has gained prominence in the last decade, but the impact of its intrinsic temporal components of DRL when applied to physical systems - generally summarized as the timestep duration -, remains underexplored. While some approaches to this issue include timestep duration as part of action space, they offer a limited analysis, overlooking key elements such as software architecture asynchrony or system-induced latencies. These factors cause deviations between nominal and actual timesteps, negatively impacting learning. This article proposes a set of navigation tasks for mobile robots specifically designed to analyze the effects of timestep duration, supported by over 2000 hours of simulation. To our knowledge, this constitutes the first multimetric evaluation of the influence of step time duration on DRL-based robotic tasks. The findings confirm the existence of optimal step times that maximize learning efficiency and task completion.
- Open Access
- 1 September 2026
- Published by Universidade da Coruna in Jornadas de Automática
- No. 47
Open-source, portable, and low-cost educational kit for engineering control courses
Hands-on experimentation is essential in control engineering education; however, traditional laboratory platforms are often costly, non-portable, or rely on proprietary software, limiting their accessibility and scalability. This paper presents an open-source, portable, and low-cost educational kit for practical learning in control engineering courses. The kit integrates an ESP32-based embedded system with a direct-current (DC) motor equipped with an encoder, together with a Python graphical user interface (GUI) for experiment configuration, real-time visualization, and data logging. The firmware employs FreeRTOS to ensure deterministic sampling and supports multiple operating modes (open-loop, position, and velocity). The GUI enables users to configure controllers, apply reference signals, and export experimental data. Thanks to its low cost and open nature, students can assemble the system themselves and have a personal experimental kit, maintaining interaction with a real physical system even outside the laboratory.
- 1 September 2026
- Published by Universidade da Coruna in Jornadas de Automática
- No. 47
Open AccessOpen-source, portable, and low-cost educational kit for engineering control courses
Hands-on experimentation is essential in control engineering education; however, traditional laboratory platforms are often costly, non-portable, or rely on proprietary software, limiting their accessibility and scalability. This paper presents an open-source, portable, and low-cost educational kit for practical learning in control engineering courses. The kit integrates an ESP32-based embedded system with a direct-current (DC) motor equipped with an encoder, together with a Python graphical user interface (GUI) for experiment configuration, real-time visualization, and data logging. The firmware employs FreeRTOS to ensure deterministic sampling and supports multiple operating modes (open-loop, position, and velocity). The GUI enables users to configure controllers, apply reference signals, and export experimental data. Thanks to its low cost and open nature, students can assemble the system themselves and have a personal experimental kit, maintaining interaction with a real physical system even outside the laboratory.
- Open Access
- 1 January 2026
- Published by Tech Science Press in Computer Modeling in Engineering & Sciences
- Vol. 148 (1) , 1-10
Modeling Time-Aware Mobile Robot Navigation by Learning Subjective Time Maps (STM)
The basic operation of a mobile robot is navigating to some target, avoiding collisions and possibly minimizing other criteria. A diversity of methods have been developed since the past century, and the research is still active, but there is one aspect that is often neglected: the duration of the steps in which computational devices divide the navigation process. Usually, it is set heuristically to a small, constant value for sampling observations frequently enough to ensure safety; however, each robot and environment has particularities that can make such a fixed timestep sub-optimal under some criteria. This paper explores the possibility of learning the best time for each step in a particular robot-environment interaction, and shows that those subjective, variable timesteps constitute an important aspect of both motion safety and efficiency. The proposed approach has an initial learning stage in a general, minimalistic environment designed for providing the main situations the robot will face; through deep RL, a time-aware reactive navigation policy is found that optimizes collision avoidance, total navigation time, and accurateness in reaching the target, yielding (motion, timestep-duration) commands. Due to the generalization capabilities of DRL, that policy induces Subjective Time Maps (STMs) for time-aware navigation in other scenarios. Thus, STMs become time-aware subjective models of the robot-world interaction that improve that particular robot navigation and also allow it to leverage time effort, e.g., longer timesteps can be used for energy savings or for other operations. Furthermore, the integration of uncertainties in both localization and decision-making is addressed. To achieve this, a time-aware motion policy is first learned while ignoring localization uncertainty, in order to improve learning convergence and reduce computational cost. The policy is then integrated with the resulting STMs through a particle filter (PF), either by using the time-aware motion command provided by the STM on the representative particle or by associating each particle (robot pose hypothesis) and its importance (weight) with it. The goal is to enable a full treatment of uncertainty with reduced complexity. Diverse experiments and comparisons with STMs show the utility of this spatio-temporal representation of knowledge for safe navigation, and the pros and cons of the two methods of uncertainty integration.
- 1 January 2026
- Published by Tech Science Press in Computer Modeling in Engineering & Sciences
- Vol. 148 (1) , 1-10
Open AccessModeling Time-Aware Mobile Robot Navigation by Learning Subjective Time Maps (STM)
The basic operation of a mobile robot is navigating to some target, avoiding collisions and possibly minimizing other criteria. A diversity of methods have been developed since the past century, and the research is still active, but there is one aspect that is often neglected: the duration of the steps in which computational devices divide the navigation process. Usually, it is set heuristically to a small, constant value for sampling observations frequently enough to ensure safety; however, each robot and environment has particularities that can make such a fixed timestep sub-optimal under some criteria. This paper explores the possibility of learning the best time for each step in a particular robot-environment interaction, and shows that those subjective, variable timesteps constitute an important aspect of both motion safety and efficiency. The proposed approach has an initial learning stage in a general, minimalistic environment designed for providing the main situations the robot will face; through deep RL, a time-aware reactive navigation policy is found that optimizes collision avoidance, total navigation time, and accurateness in reaching the target, yielding (motion, timestep-duration) commands. Due to the generalization capabilities of DRL, that policy induces Subjective Time Maps (STMs) for time-aware navigation in other scenarios. Thus, STMs become time-aware subjective models of the robot-world interaction that improve that particular robot navigation and also allow it to leverage time effort, e.g., longer timesteps can be used for energy savings or for other operations. Furthermore, the integration of uncertainties in both localization and decision-making is addressed. To achieve this, a time-aware motion policy is first learned while ignoring localization uncertainty, in order to improve learning convergence and reduce computational cost. The policy is then integrated with the resulting STMs through a particle filter (PF), either by using the time-aware motion command provided by the STM on the representative particle or by associating each particle (robot pose hypothesis) and its importance (weight) with it. The goal is to enable a full treatment of uncertainty with reduced complexity. Diverse experiments and comparisons with STMs show the utility of this spatio-temporal representation of knowledge for safe navigation, and the pros and cons of the two methods of uncertainty integration.
- Open Access
Goodness of Fit in the Marginal Modeling of Round-Trip Times for Networked Robot Sensor Transmissions
When complex computations cannot be performed on board a mobile robot, sensory data must be transmitted to a remote station to be processed, and the resulting actions must be sent back to the robot to execute, forming a repeating cycle. This involves stochastic round-trip times in the case of non-deterministic network communications and/or non-hard real-time software. Since robots need to react within strict time constraints, modeling these round-trip times becomes essential for many tasks. Modern approaches for modeling sequences of data are mostly based on time-series forecasting techniques, which impose a computational cost that may be prohibitive for real-time operation, do not consider all the delay sources existing in the sw/hw system, or do not work fully online, i.e., within the time of the current round-trip. Marginal probabilistic models, on the other hand, often have a lower cost, since they discard temporal dependencies between successive measurements of round-trip times, a suitable approximation when regime changes are properly handled given the typically stationary nature of these round-trip times. In this paper we focus on the hypothesis tests needed for marginal modeling of the round-trip times in remotely operated robotic systems with the presence of abrupt changes in regimes. We analyze in depth three common models, namely Log-logistic, Log-normal, and Exponential, and propose some modifications of parameter estimators for them and new thresholds for well-known goodness-of-fit tests, which are aimed at the particularities of our setting. We then evaluate our proposal on a dataset gathered from a variety of networked robot scenarios, both real and simulated; through >2100 h of high-performance computer processing, we assess the statistical robustness and practical suitability of these methods for these kinds of robotic applications.
Open AccessGoodness of Fit in the Marginal Modeling of Round-Trip Times for Networked Robot Sensor Transmissions
When complex computations cannot be performed on board a mobile robot, sensory data must be transmitted to a remote station to be processed, and the resulting actions must be sent back to the robot to execute, forming a repeating cycle. This involves stochastic round-trip times in the case of non-deterministic network communications and/or non-hard real-time software. Since robots need to react within strict time constraints, modeling these round-trip times becomes essential for many tasks. Modern approaches for modeling sequences of data are mostly based on time-series forecasting techniques, which impose a computational cost that may be prohibitive for real-time operation, do not consider all the delay sources existing in the sw/hw system, or do not work fully online, i.e., within the time of the current round-trip. Marginal probabilistic models, on the other hand, often have a lower cost, since they discard temporal dependencies between successive measurements of round-trip times, a suitable approximation when regime changes are properly handled given the typically stationary nature of these round-trip times. In this paper we focus on the hypothesis tests needed for marginal modeling of the round-trip times in remotely operated robotic systems with the presence of abrupt changes in regimes. We analyze in depth three common models, namely Log-logistic, Log-normal, and Exponential, and propose some modifications of parameter estimators for them and new thresholds for well-known goodness-of-fit tests, which are aimed at the particularities of our setting. We then evaluate our proposal on a dataset gathered from a variety of networked robot scenarios, both real and simulated; through >2100 h of high-performance computer processing, we assess the statistical robustness and practical suitability of these methods for these kinds of robotic applications.
- Open Access
- 12 July 2024
- Published by Universidade da Coruna in Jornadas de Automática
- No. 45
Sistema multi-cámara de estimación de pose sin marcadores para pHRI
Este trabajo presenta un sistema basado en visión que utiliza redes neuronales para la estimación de poses humanas en 3D. La solución desarrollada identifica en el sujeto analizado 18 "puntos clave" o keypoints mediante cuatro cámaras RGB calibradas. La utilización de múltiples cámaras permite superar problemas inherentes al uso de una sola cámara RGBD/estéreo como la pérdida de keypoints por la existencia de oclusiones o una mayor incertidumbre en la estimación de la profundidad, proporcionando una base robusta para futuras investigaciones y aplicaciones en campos como la rehabilitación física. Asimismo, se presenta un dataset, que ha sido puesto a disposición de la comunidad, con la posición 3D de los keypoints identificados durante la realización de seis ejercicios distintos. Este trabajo contribuye a la literatura actual ofreciendo un enfoque novedoso en la recopilación y análisis de datos de posturas humanas y demuestra la viabilidad de nuestra metodología, abriendo nuevas vías para investigaciones futuras en el contexto del pHRI.
- 12 July 2024
- Published by Universidade da Coruna in Jornadas de Automática
- No. 45
Open AccessSistema multi-cámara de estimación de pose sin marcadores para pHRI
Este trabajo presenta un sistema basado en visión que utiliza redes neuronales para la estimación de poses humanas en 3D. La solución desarrollada identifica en el sujeto analizado 18 "puntos clave" o keypoints mediante cuatro cámaras RGB calibradas. La utilización de múltiples cámaras permite superar problemas inherentes al uso de una sola cámara RGBD/estéreo como la pérdida de keypoints por la existencia de oclusiones o una mayor incertidumbre en la estimación de la profundidad, proporcionando una base robusta para futuras investigaciones y aplicaciones en campos como la rehabilitación física. Asimismo, se presenta un dataset, que ha sido puesto a disposición de la comunidad, con la posición 3D de los keypoints identificados durante la realización de seis ejercicios distintos. Este trabajo contribuye a la literatura actual ofreciendo un enfoque novedoso en la recopilación y análisis de datos de posturas humanas y demuestra la viabilidad de nuestra metodología, abriendo nuevas vías para investigaciones futuras en el contexto del pHRI.
- Open Access
- 12 July 2024
- Published by Universidade da Coruna in Jornadas de Automática
- No. 45
Aprendizaje por refuerzo en vivo para robots móviles
En este trabajo se propone una arquitectura de reflejos ante estímulos sensoriales para que el aprendizaje por refuerzo en vivo para robots móviles mejore su adaptación a cambios en la tarea, aumente su autonomía para regresar a estados seguros tras errores y reduzca, en general, la supervisión por parte del humano. El trabajo se ha enfocado en la navegación de un robot móvil con evitación de obstáculos y hemos utilizado una versión modificada de los algoritmos de aprendizaje por refuerzo Q-learning y True On-Line SARSA (lambda). Se ha estudiado adicionalmente un aprendizaje que traslada lo aprendido en simulación al aprendizaje en vivo, llamado aprendizaje híbrido. Los resultados muestran que nuestra arquitectura mejora la seguridad del robot y su adaptabilidad a cambios en la tarea, minimiza la intervención humana y extiende el tiempo de entrenamiento sin supervisión.
- 12 July 2024
- Published by Universidade da Coruna in Jornadas de Automática
- No. 45
Open AccessAprendizaje por refuerzo en vivo para robots móviles
En este trabajo se propone una arquitectura de reflejos ante estímulos sensoriales para que el aprendizaje por refuerzo en vivo para robots móviles mejore su adaptación a cambios en la tarea, aumente su autonomía para regresar a estados seguros tras errores y reduzca, en general, la supervisión por parte del humano. El trabajo se ha enfocado en la navegación de un robot móvil con evitación de obstáculos y hemos utilizado una versión modificada de los algoritmos de aprendizaje por refuerzo Q-learning y True On-Line SARSA (lambda). Se ha estudiado adicionalmente un aprendizaje que traslada lo aprendido en simulación al aprendizaje en vivo, llamado aprendizaje híbrido. Los resultados muestran que nuestra arquitectura mejora la seguridad del robot y su adaptabilidad a cambios en la tarea, minimiza la intervención humana y extiende el tiempo de entrenamiento sin supervisión.
- Open Access
- 11 July 2023
- Published by Elsevier in Engineering Applications of Artificial Intelligence
- Vol. 125
A multidimensional Bayesian architecture for real-time anomaly detection and recovery in mobile robot sensory systems
- 11 July 2023
- Published by Elsevier in Engineering Applications of Artificial Intelligence
- Vol. 125
Open AccessA multidimensional Bayesian architecture for real-time anomaly detection and recovery in mobile robot sensory systems
- 1 November 2020
- Published by IATED Academy in ICERI Proceedings
- Vol. 1, 443-450
OPTIMIZING SUBJECT DESIGN, TIMING, AND FOCUS IN A DIVERSITY OF ENGINEERING COURSES THROUGH THE USE OF A LOW-COST ARDUINO SHIELD
Information about the paper titled "OPTIMIZING SUBJECT DESIGN, TIMING, AND FOCUS IN A DIVERSITY OF ENGINEERING COURSES THROUGH THE USE OF A LOW-COST ARDUINO SHIELD" at IATED Digital Library
- 1 November 2020
- Published by IATED Academy in ICERI Proceedings
- Vol. 1, 443-450
OPTIMIZING SUBJECT DESIGN, TIMING, AND FOCUS IN A DIVERSITY OF ENGINEERING COURSES THROUGH THE USE OF A LOW-COST ARDUINO SHIELD
Information about the paper titled "OPTIMIZING SUBJECT DESIGN, TIMING, AND FOCUS IN A DIVERSITY OF ENGINEERING COURSES THROUGH THE USE OF A LOW-COST ARDUINO SHIELD" at IATED Digital Library
- Open Access
- 7 April 2020
- Published by Universitat Politecnica de Valencia in Revista Iberoamericana de Automática e Informática Industrial
- Vol. 17 (2) , 107-115
Minivídeos tutoriales como apoyo al aprendizaje de conceptos básicos para un curso de Fundamentos de Control Automático
En este trabajo se presenta una iniciativa encaminada a introducir y fomentar el uso de los videotutoriales como herramienta de apoyo para el aprendizaje de los conceptos básicos del control automático en titulaciones universitarias técnicas. Los contenidos desarrollados, disponibles en el canal de YouTube denominado “Automática”, se han agrupado de acuerdo a los principales bloques conceptuales de la materia y han sido puestos a disposición del alumnado en el espacio del campus virtual de la Universidad de Málaga (UMA) asociado a 13 asignaturas impartidas por el Dpto. de Ingeniería de Sistemas y Automática de la UMA en los cursos académicos 2017/2018 y 2018/2019. Los resultados obtenidos a través de encuestas de opinión cumplimentadas por los estudiantes muestran la buena acogida de los mismos. Por otro lado, el cruce de los datos de uso (entendido como tiempo efectivo de reproducción) recogidos durante este periodo y las calificaciones obtenidas a través de pruebas de conocimiento diseñadas ad-hoc revelan la existencia de una relación entre el empleo del material desarrollado y la mejora del rendimiento académico de los estudiantes.
- 7 April 2020
- Published by Universitat Politecnica de Valencia in Revista Iberoamericana de Automática e Informática Industrial
- Vol. 17 (2) , 107-115
Open AccessMinivídeos tutoriales como apoyo al aprendizaje de conceptos básicos para un curso de Fundamentos de Control Automático
En este trabajo se presenta una iniciativa encaminada a introducir y fomentar el uso de los videotutoriales como herramienta de apoyo para el aprendizaje de los conceptos básicos del control automático en titulaciones universitarias técnicas. Los contenidos desarrollados, disponibles en el canal de YouTube denominado “Automática”, se han agrupado de acuerdo a los principales bloques conceptuales de la materia y han sido puestos a disposición del alumnado en el espacio del campus virtual de la Universidad de Málaga (UMA) asociado a 13 asignaturas impartidas por el Dpto. de Ingeniería de Sistemas y Automática de la UMA en los cursos académicos 2017/2018 y 2018/2019. Los resultados obtenidos a través de encuestas de opinión cumplimentadas por los estudiantes muestran la buena acogida de los mismos. Por otro lado, el cruce de los datos de uso (entendido como tiempo efectivo de reproducción) recogidos durante este periodo y las calificaciones obtenidas a través de pruebas de conocimiento diseñadas ad-hoc revelan la existencia de una relación entre el empleo del material desarrollado y la mejora del rendimiento académico de los estudiantes.