I am a Computer and Mechatronics Engineer and current Research Associate and PhD student at the Machine Perception and Intelligent Robotics (MAPIR-UMA) group, University of Málaga, Spain. My work focuses on studying and developing Computer Vision and Artificial Intelligence methods to enhance the cognitive capabilities of mobile robots.

I combine solid academic training with practical software engineering experience.
My background includes:

• B.Sc. in Computer Engineering (University of Málaga, 2024).
• Erasmus stay at the University of Padua, specializing in Computer Vision, Robotics, and Natural Language Processing.
• Master’s in Mechatronics Engineering (University of Málaga, 2025).
• Fluent in Spanish, English, and Italian.

My work is driven by a deep passion for bringing advanced AI technologies into real-world applications that make everyday life better, from empowering mobile robots with richer perception and reasoning to creating intelligent systems that seamlessly support people in their daily activities.

• ImageColouring (2022) – Final project for the Neural Networks and Deep Learning course. A conditional GAN project using U-Net and PatchGAN to recolor grayscale images. It leverages LPIPS perceptual loss and a WGAN discriminator with gradient penalty to boost color realism and maintain stable training, achieving visually convincing results on a toy dataset. This work deepened my understanding of deep learning architectures, loss functions, and training stability techniques.

• tiago-pick-place-ir (2022) – Final project for the Intelligent Robotics course. Developed a set of ROS 2 nodes to perform a complete pick-and-place task with a simulated TIAGo mobile manipulator. The system navigates through the room, reaches designated locations, and controls the robot arm to grasp and place objects from a table, integrating navigation, perception, and manipulation in a single pipeline. This project strengthened my skills in ROS 2 development, robot motion planning, and the integration of navigation and manipulation tasks in mobile robotics.

• malaga-bus-graph (2023) – Final project for the Learning from Networks course. Conducted a graph-based analysis of Málaga's public bus transport system to study frequency, speed, and node centralities. Built a large, directed multi-graph of bus lines, stops, routes, and timetables, then applied closeness and betweenness centrality metrics and compared them with random graph models to assess significance. This project enhanced my understanding of network analysis, graph algorithms, and their practical application to real-world transportation data.

• neural-dependency-parser-spanish (2023) – Final project for the Natural Language Processing course. Implementation of a neural transition-based dependency parser for Spanish using the arc-eager algorithm and a static oracle, featuring both a Bi-LSTM baseline and a BERT-based (BETO) model. The system is trained and evaluated on the Universal Dependencies es_AnCora treebank and includes full data preprocessing, parser/oracle design, and training–evaluation pipelines built with PyTorch and Hugging Face.

• semantics-maps-moncada (2024) – Code for my B.Sc. final thesis "Building and Exploiting Semantic Maps in Robotics Using Large Models". Implements an end-to-end pipeline, based on ConceptGraphs, that generates open-vocabulary semantic maps by combining Segment Anything (SAM), CLIP, Gemini, and ChatGPT to detect objects and infer their relationships. The system produces scene-graph representations validated on Replica and ScanNet datasets and was the basis for the paper "Modelos a gran escala para mapeo semántico en robótica móvil" (Large models for semantic mapping in mobile robotics) presented at Jornadas de Automática 2024.

• Metacontratas – Contributed as a backend developer, focusing on authentication and secure access control and on enhancing the document validation pipeline.

• llm-robotics-reflection (2025) – Research project and codebase for the paper Agentic Workflows for Improving LLM Reasoning in Robotic Object-Centered Planning, published in MDPI Robotics. Provides the dataset, prompts, and Python implementation to evaluate agentic workflows—Self-Reflection, Multi-Agent Reflection, and LLM Ensemble— for improving Large Language Model reasoning in robotic object-centered planning. Experiments on semantic maps from ScanNet and SceneNN demonstrate significant gains in object retrieval accuracy, strengthening my skills in LLM prompting, agentic workflow design, and advanced AI–robotics integration.

• generative-topological-maps (2025) – Framework for place segmentation, place categorization, and semantic object-relationship inference in pre-built 3D semantic maps. Developed as my Master’s Thesis in Mechatronics Engineering at the University of Málaga, this work introduces a geometric–semantic clustering pipeline and LLM/LVLM methods to create rich semantic-topological maps for mobile robots. Key results were published in the paper Mobile Robot Place Segmentation and Categorization Using Object Semantics, presented at the European Conference on Mobile Robots (ECMR) 2025, and fully detailed in my Master’s Thesis report. This project deepened my expertise in geometric–semantic mapping, LLM/LVLM integration, and the design of advanced pipelines for robotic reasoning.

• rol_lm (2025, under active development) – ROS 2 package designed to connect robotic systems with LLMs and Large Vision-Language Models (LVLMs) through a simple, unified service. The project provides an accessible way for robots to request text or vision-based reasoning from powerful generative models, making it easier to integrate state-of-the-art AI into real-world robotic applications. Working on this repository strengthened my skills in ROS 2 architecture, service design, and the practical integration of modern AI models within robotic platforms.

• ScanNet (2024, under development) – Utilities to automatically download ScanNet sequences and create ready-to-use ROS1/ROS2 bag files for robotic experiments.

• SceneNN (2024, under development) – Companion utilities for downloading SceneNN dataset sequences and generating ROS-compatible bag files for easy integration in robotics projects.

• Email: jemonra@gmail.com
• LinkedIn: linkedin.com/in/jesús-moncada-ramírez-3a79971a4
• Website/Portfolio: Coming soon