PFPablo Malvido Fresnillo

Pablo Malvido Fresnillo

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Pablo Malvido Fresnillo has a total of 21 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Vision-Guided Robotic System for Automatic Fish Quality Grading and Packaging

    by Mohamed Lamine Mekhalfi, Saigopal Vasudevan, Jorge S. Calado, Anh Dong Le, Pablo Malvido Fresnillo, Jose Ferreira, Pedro Garcia, Carla Macieira, Paul Ian Chippendale, Ricardo Jardim-Gonçalves, et al.

    Dear Editor, Automated fish grading and packaging, critical in the seafood industry, have not been simultaneously addressed by prior works using machine vision and robotics. This letter presents a novel proof-of-concept robotic vision system for automatic, size-based fish grading and packaging. Our system classifies frozen fish steaks into two size grades and localizes them on a conveyor belt for robotic pick-and-place via a specialized end-effector. Experiments achieved a grading accuracy of 87.6% and a robotic packaging rate of 87%, demonstrating the potential of vision-guided robotics for automated food quality inspection and handling.


  • Efficient Dataset Generation for Stacked Meat Products Instance Segmentation in Food Automation

    by Hoang Minh Pham, Anh Dong Le, Pablo Malvido-Fresnillo, Saigopal Vasudevan, José L. Martínez Lastra

    Dear Editor, This letter presents techniques to simplify dataset generation for instance segmentation of raw meat products, a critical step toward automating food production lines. Accurate segmentation is essential for addressing challenges such as occlusions, indistinct edges, and stacked configurations, which demand large, diverse datasets. To meet these demands, we propose two complementary approaches: a semi-automatic annotation interface using tools like the segment anything model (SAM) and GrabCut and a synthetic data generation pipeline leveraging 3D-scanned models. These methods reduce reliance on real meat, mitigate food waste, and improve scalability. Experimental results demonstrate that incorporating synthetic data enhances segmentation model performance and, when combined with real data, further boosts accuracy, paving the way for more efficient automation in the food industry.

  • Open Access

    A dual-arm robotic system for automated multi-branch wire harness assembly in automotive industry

    by Pablo Malvido Fresnillo, Saigopal Vasudevan, Wael M. Mohammed, Jose A. Perez Garcia, Jose L. Martinez Lastra


  • Open Access

    Bimanual Motion Primitives for Manipulating 3D Deformable Objects Having Biological Tissues on a Planar Surface With Passive End-Effectors

    by Hoang Pham, Pablo Malvido Fresnillo, Saigopal Vasudevan, Wael M. Mohammed, Jose L. Martinez Lastra

    Manipulating deformable objects (DOs) that change shape during operations poses a challenge for robots due to the complexity of modelling and predicting the object’s state. This challenge is critical for the food industry, where collaborative robots will assist workers in processing and packaging food products. Most meat products are three-dimensional DOs (3D-DOs) composed of biological tissues, giving them visco-elastic and anisotropic properties that make their behaviour highly unpredictable and challenging to model. To address this challenge and to minimise food cross-contamination, dual-arm robots and non-actuated end-effectors such as chopsticks are used to provide more flexibility, especially when working with objects in confined spaces and reducing tool actuation time. Our goal is to develop a methodology for manipulating predominant convex 3D-DOs composed of biological tissues on a planar surface, altering their position and orientation while minimizing their deformations. To accomplish this, we propose bimanual motion primitives utilising a novel geometric approach. Effectiveness is measured by the Intersection-over-Union ratio and object centroid error metrics through 180 chicken-breast, 36 pork-loin-steak experiments, and 96 comparative experiments with heuristic methods for rigid objects.

  • Open Access

    Machine Vision and Robotics for Primary Food Manipulation and Packaging: A Survey

    by Saigopal Vasudevan, Mohamed Lamine Mekhalfi, Carlos Blanes, Michela Lecca, Fabio Poiesi, Paul Ian Chippendale, Pablo Malvido Fresnillo, Wael M. Mohammed, Jose L. Martinez Lastra

    Robotic and vision technologies are progressively becoming ubiquitous for automating and digitizing quality control in the food industry. This paper aspires to provide a high-level technical review on the crucial role of advanced automation technologies, including versatile or dedicated robotic systems, specialized end-effectors, machine vision, and efficient material handling systems, which collectively enhance food processing efficiency. While the manuscript aims to document the various automation sub-systems utilized generally in food processing, it places a particular emphasis on the primary processing phase of food production. Most food products in the primary processing phase exhibit a plethora of complex physical properties and manipulation conditions, making it difficult to reliably automate the various processes. This research aims to outline the contemporary advances and requirements for integrating various automation technologies, to enhance the efficiency and precision of primary food processing. Furthermore, it aspires to serve as a valuable, up-to-date survey and analysis of the latest advances in automation and vision technologies and their capability to automate a food processing line.


  • Open Access

    An Open and Reconfigurable User Interface to Manage Complex ROS-Based Robotic Systems

    by Pablo Malvido Fresnillo, Saigopal Vasudevan, Jose A. Perez Garcia, Jose L. Martinez Lastra

    The Robot Operating System (ROS) has significantly gained popularity among robotic engineers and researchers over the past five years, primarily due to its powerful infrastructure for node communication, which enables developers to build modular and large robotic applications. However, ROS presents a steep learning curve and lacks the intuitive usability of vendor-specific robotic Graphical User Interfaces (GUIs). Moreover, its modular and distributed nature complicates the control and monitoring of extensive systems, even for advanced users. To address these challenges, this paper introduces a novel, highly adaptable and reconfigurable web-based GUI for intuitively controlling, monitoring, and configuring complex ROS-based robotic systems. The GUI leverages ROSBridge and roslibjs to ensure seamless communication with ROS systems via topics and services. Special attention has been paid to enable the easy reconfiguration of the GUI and to minimize the modifications required to integrate it with new robotic systems. Thus, it has been designed as a versatile platform, which allows for the selective incorporation of modular features to accommodate diverse robotic systems and applications. A predefined template has been defined to ensure the compatibility of the developed features with the GUI platform. Additionally, an initial set of commonly used features in robotic applications is presented. To demonstrate its reconfigurability, the GUI was customized and tested for four industrial use cases, receiving positive feedback. The project’s repository has been made publicly available to support the robotics community and lower the entry barrier for ROS in industrial applications.


  • Open Access

    Generation of realistic synthetic cable images to train deep learning segmentation models

    by Pablo MalvidoFresnillo, Wael M. Mohammed, Saigopal Vasudevan, Jose A. PerezGarcia, Jose L. MartinezLastra

    Semantic segmentation is one of the most important and studied problems in machine vision, which has been solved with high accuracy by many deep learning models. However, all these models present a significant drawback, they require large and diverse datasets to be trained. Gathering and annotating all these images manually would be extremely time-consuming, hence, numerous researchers have proposed approaches to facilitate or automate the process. Nevertheless, when the objects to be segmented are deformable, such as cables, the automation of this process becomes more challenging, as the dataset needs to represent their high diversity of shapes while keeping a high level of realism, and none of the existing solutions have been able to address it effectively. Therefore, this paper proposes a novel methodology to automatically generate highly realistic synthetic datasets of cables for training deep learning models in image segmentation tasks. This methodology utilizes Blender to create photo-realistic cable scenes and a Python pipeline to introduce random variations and natural deformations. To prove its performance, a dataset composed of 25000 synthetic cable images and their corresponding masks was generated and used to train six popular deep learning segmentation models. These models were then utilized to segment real cable images achieving outstanding results (over 70% IoU and 80% Dice coefficient for all the models). Both the methodology and the generated dataset are publicly available in the project’s repository.


  • Open Access

    Generation of realistic synthetic cable images to train Deep Learning segmentation models

    by Pablo Malvido Fresnillo, Wael M. Mohammed, Saigopal Vasudevan, Jose A. Perez Garcia, Jose L. Martinez Lastra

    One of the main contemporary challenges in robotics is the automation of tasks that involve handling Deformable Linear Objects (DLOs), such as cables or ropes. Due to their changing shape, constant perception is required, which makes computer vision, and in particular, image segmentation, fundamental. Even though image segmentation is a very studied problem, which has been addressed with high accuracy by many different Deep Learning models, they need to be properly trained to segment the objects of interest (DLOs in this case). To do this, it is necessary to have a large and diverse dataset. However, such a dataset hasn’t been created for cable images, and doing it manually would be complicated and extremely time-consuming. This paper addresses this issue with a novel methodology, which automatically generates synthetic cable datasets to train image segmentation models. This methodology utilizes Blender to create photo-realistic scenes and a Python pipeline to interact with them. To ensure the diversity of the dataset, before the generation of each image, the pipeline performs random variations on the elements of the scene. After this, the scene is animated, dropping the cables from a certain height and letting them deform naturally after landing, which results in realistic arrangements of the cables. The effectiveness of the methodology was demonstrated by training six popular segmentation models with synthetic datasets and using them to segment real cable images with great results (IoU over 70% and Dice coefficient over 80% for all the models).

  • Open Access

    An approach based on machine vision for the identification and shape estimation of deformable linear objects

    by Pablo Malvido Fresnillo, Saigopal Vasudevan, Wael M. Mohammed, Jose L. Martinez Lastra, Jose A. Perez Garcia


  • Open Access

    A method for understanding and digitizing manipulation activities using programming by demonstration in robotic applications

    by Pablo Malvido Fresnillo, Saigopal Vasudevan, Wael M. Mohammed, Jose L. Martinez Lastra, José A. Pérez García