Oportunidades Públicas

Reconstruction of incomplete images

The images submuestreadas are a way to increase the speed of acquisition for the magnetic resonance imaging. In this project we will study different spaces in which they can represent the data in a manner to facilitate its reconstruction by employing algorithms of Compressed Sensing.

Magnetic Resonance Imaging Acquisition and Reconstruction

Deep learning based super resolution reconstruction for high resolution coronary magnetic resonance angiography

Coronary artery disease is currently diagnosed with X-ray or computed tomography coronary angiography (CTCA). However, both modalities expose patients to ionising radiation and iodinated contrast agents (CA). We have developed an alternative approach, coronary magnetic resonance angiography (CMRA), which is non-invasive, does not require the injection of CA and is free of radiation. Our 3D CMRA technique is approaching the resolution of CTCA but still requires a 10-minute scan. To further shorten the scan time to <1 minute and match the resolution of CTCA (0.6mm3), we propose to investigate novel deep learning based super resolution (SR) image reconstruction networks. The work will include a systematic literature review and the implementation of the most promising SR networks. We have a large clinical dataset of CMRA/CTCA data that can be used for training, testing and validation. It would be useful to know about the fundamentals of artificial intelligence.

Acquisition, reconstruction and motion correction for 3D whole-heart Magnetic Resonance Imaging

Magnetic Resonance Imaging (MRI) is an important non-invasive tool for risk assessment and treatment monitoring of cardiovascular disease. In comparison to other imaging modalities (ultrasound, x-ray and computed tomography) MRI has the advantage of combining excellent soft tissue contrast with high spatial resolution. However, a major limitation of current MRI protocols is that all imaging sequences are acquired sequentially, with different resolution, geometric orientations and breath-hold positions, resulting in long planning and scan times. Moreover, image quality degradation due to respiratory and cardiac motion are major challenges that greatly affect the accuracy and reproducibility of cardiac MRI. This research includes the development of novel whole-heart 3D MRI sequences, motion compensation techniques and deep learning methods to enable 3D, easy-to-plan, fast and efficient quantitative multi-parametric MRI of heart and vessels. The research is done together with King’s College London. Basic knowledge of MRI, signal processing and artificial intelligence would be useful.

Computational creativity in generative networks for audio

Implementation of new Deep Learning architectures for Quantitative Susceptibility Mapping

We look to implement new networks like Kolmogorov-Arnold Networks, transformers, conditional flow matching and others, to improve the quality of the reconstructions in QSM

Electrification and Restoration Study of the UC East Campus Organ

The UC Campus Oriente organ, a historic instrument whose original pneumatic action has been completely destroyed, poses a technical and heritage challenge. This project seeks to conduct a preliminary feasibility study for its electrification, exploring the use of solenoids to open the secret valves that allow air to pass through, replacing the damaged original mechanisms. The study will include a technical analysis of possible solutions and the conceptual design of the electrical system, with a view to restoring its functionality and opening new possibilities for technological and musical experimentation. This research represents a unique opportunity to integrate electrical engineering knowledge with the preservation and modernization of an emblematic musical instrument. Professors: Rodrigo Cádiz (Engineering and Music), Patricio de la Cuadra (Engineering and Music), José Manuel Izquierdo (Music).

Migración de Sensores de ROS 1 a ROS 2

El sistema operativo para robots (ROS 1) ha llegado oficialmente al final de su ciclo de vida, y su sucesor, ROS 2, resuelve muchas de sus limitaciones relacionadas con el rendimiento, la seguridad y la escalabilidad. Sin embargo, muchos repositorios existentes permanecen desactualizados, lo que restringe la integración de sensores y plataformas robóticas comúnmente utilizadas. Este proyecto de IPre tiene como objetivo migrar y actualizar los controladores y repositorios de ROS 1 a ROS 2 para sensores ampliamente utilizados como el LiDAR Ouster OS1, Livox AVIA, VectorNav VN200, RTK NavCom y RTK Ardusimple. El objetivo es mejorar la compatibilidad y asegurar su uso continuo en sistemas robóticos modernos.

Embedded Deep Learning and Federated Learning for Glucose Control

In this IPRE, the student will implement Deep Learning algorithms on microcontrollers (Jetson Nano) with the aim of developing glucose control strategies. Additionally, the use of Federated Learning algorithms will be explored to coordinate the training of multiple agents, enabling more efficient learning without the need to share sensitive data.

Generative network of musical exercises for auditory training

In this research, we will attempt to train a generative deep learning model based on a database of music theory and auditory training exercises, using multimodal models that receive text in addition to scores. The idea is to implement a tool for personal study without the need for a teacher as a mediator.

An experimental platform for the control of autonomous vehicles

This research falls within the field of automatic control of IoT systems composed of autonomous vehicles that communicate with each other through wireless channels. The undergraduate research project focuses on the development and refinement of a small-scale platform for platooning autonomous vehicles using robots. Various experiments are to be implemented on this platform, including event-based control, analysis of the impact of transmission power and rate on platoon stability, and strategies to compensate for data loss, among others.

Electricity Systems Planning and Public Policy

Research on the use and development of planning models for power systems and the design of public policies. In particular, the SWITCH model for generation and transmission expansion will be applied to the Chilean power system. The impact of integrating new technologies into power systems will be quantified, along with other elements such as increasing the use of electricity in the energy matrix and leveraging demand-side flexibility, among others.

Teaching platform programming for power converters

Programming a teaching platform for power converters in applications such as PV, BESS, active filters, grid-tie, etc.

The project involves reprogramming and adapting a kit already implemented on a platform focused on educational activities. The main hardware (controller and converter) is available and functioning, but the control system does not need to be adapted for new applications.

The LAUNCHXL-F28379D microcontroller is programmed using the Target Support Package (an extension of PLECS Blockset and PLECS Standalone that allows generating code for microcontrollers).

Control platform: LAUNCHXL-F28379D
Three-phase inverter: BOOSTXL-3PhGaNInv (can also be used as a DC-DC inverter)
Motor: BLWS232D-24V-4000 (only if using drives)
Other possible applications: DC-DC for PV, BESS, grid-tie active filter, DAB, etc.

Propulsion of Ships Powered by Batteries and Green Hydrogen

The project was awarded under the 2025 Regular Technology Rings Competition of the National Research and Development Agency (ANID). The project aims to develop a comprehensive framework for the efficient management of electric vessels that use batteries and hydrogen as energy sources. The initiative aligns with the Chilean Navy's National Green Hydrogen Strategy and its Continuous National Shipbuilding Plan, aiming to strengthen the sustainability and competitiveness of the national maritime sector. The project also has the support of the Chilean Navy and the companies Saam Towage and Dhemax, which reinforces its applied focus and its potential impact on national industry.

Main lines of research:

1. Optimal management of energy vectors under uncertainty.

2. The second focuses on electrical infrastructure and microgrids.

3. Electric propulsion and conversion systems.

IPRE is focused on line 3 but has potential for collaboration in the other lines.

Evaluation of Multifrequency Deep Image Prior for Real-Time 0.55 T Radial Cardiac Cine MRI

Cardiac cine Magnetic Resonance Imaging (MRI) is the gold standard for assessing heart function, but conventional techniques rely on electrocardiogram(ECG)-based acquisitions over multiple heartbeats, which limits temporal resolution and excludes patients with irregular heart rhythms. Recent advances in artificial intelligence have introduced Time-Dependent Deep Image Prior (TD-DIP), a powerful method that enables high–temporal-resolution cardiac imaging from very short MRI acquisitions. However, standard TD-DIP still depends on ECG signals, restricting its applicability in patients with arrhythmias. The Multifrequency TD-DIP framework overcomes this limitation by learning cardiac motion directly from the data—without requiring ECG input. This method has shown promising results at 1.5 T using spiral acquisitions. The objective of this project is to implement and evaluate Multifrequency TD-DIP for radial cardiac cine MRI at 0.55 T. Low-field MRI systems offer improved patient comfort, lower costs, and greater accessibility, but they also present exciting challenges due to reduced signal-to-noise ratio (SNR). This project offers hands-on experience at the intersection of MRI physics, signal processing, and deep learning, contributing to more accessible and patient-friendly cardiac MR imaging technologies.

Deep Learning–Based T1 and T2 3D Whole-Heart Mapping for Joint T1/T2 MRI

T1 and T2 mapping are cardiac Magnetic Resonance Imaging (MRI) techniques that have been shown to provide objective information about the myocardium, enabling the assessment of cardiac diseases that cannot be diagnosed using conventional cardiac imaging. Joint T1/T2 sequence enables the acquisition of simultaneous 3D T1 and T2 mapping from 3 water volumes. Since T1 and T2 maps are now estimated in 3D, this process can be time-consuming. The goal of this project is to train a neural network to directly estimate T1 and T2 maps from a novel joint T1/T2 sequence at 0.55T, reducing processing time and improving its accuracy.

SCORPIO: Satellite COmmunication and Radio Propagation for the Internet Of Things

SCORPIO: Satellite COmmunication and Radio Propagation for the Internet Of Things is a research project aimed at characterizing wireless signal propagation in direct-to-satellite IoT systems, with a particular focus on LoRa-based technologies operating in the VHF/UHF bands. The project involves the design and deployment, at the UC Department of Electrical Engineering, of a ground station for satellite beacon reception, the acquisition of metrics such as RSSI and SNR, and the subsequent empirical analysis and modeling of the radio channel. The objective is to develop realistic propagation models that enable reliable evaluation and design of Sat-IoT networks under real operating conditions, taking into account Chile’s geographical diversity.

To carry out the project, we seek students with interest in wireless communications or radio-frequency systems and, ideally, with advanced knowledge of microcontrollers, SDR platforms, and embedded systems in general. Beyond satellite beacon reception, the project requires contributions in three key areas: power supply, to design a solution that reliably powers the entire system from the electrical grid, accounting for different voltage requirements across components and the need for backup mechanisms during power outages; data management, to define and maintain a database that supports efficient storage and analysis of the collected data; and graphical interface development, to create a modern, integrated visualization framework that facilitates data interpretation. This IPRE is particularly well suited for students motivated by hands-on, interdisciplinary experimental projects.

ISAC: Integrated Sensing and Communication using Wi-Fi Channel State Information

ISAC: Integrated Sensing and Communication using Wi-Fi Channel State Information is an IPRE focused on the use of Channel State Information (CSI) from Wi-Fi links to jointly support communication and sensing functionalities. Leveraging the Linux 802.11n CSI Tool, the project aims to study how fine-grained channel measurements extracted at the physical layer can be exploited not only to analyze wireless link behavior, but also to infer properties of the surrounding environment. By analyzing temporal and spectral variations of CSI, the project explores ISAC use cases such as motion detection, environmental changes, or geometry-related effects, using standard Wi-Fi hardware without dedicated sensing devices.

To carry out the project, we seek students with interest in wireless communications, radio propagation, or signal processing, and preferably with experience in Linux-based systems, low-level networking, and embedded or SDR platforms. The work involves configuring and operating the Linux 802.11n CSI Tool, designing controlled experiments for CSI data collection, and developing signal processing and data analysis pipelines for sensing applications. Depending on the student’s profile, the project may emphasize experimental setup, CSI preprocessing and analysis, or the development of prototype ISAC applications. This IPRE is particularly well suited for students interested in hands-on experimentation and research at the intersection of wireless communications and sensing.

Reinforcement Learning for the Optimization of Cross-Training Workout Plans

In this IPRE, the student will develop Reinforcement Learning algorithms aimed at the optimization of personalized training plans for cross-training or CrossFit-style disciplines. The goal is to design training strategies that balance progression, intensity, and recovery, taking into account the athlete’s physiological adaptation and the prevention of overtraining.

As a first step, a dynamic model of the athlete will be developed, describing the evolution of variables such as fitness, fatigue, and performance as a function of training load. This model will serve as the basis for formulating the problem as a sequential control system, in which a reinforcement learning agent acts as an intelligent coach that selects training sessions over time.

The project will be approached from an intelligent control perspective, exploring both model-free and model-based methods, and validating the results through simulations.

Diffussion Language Models for Music Generation

The goal of this research opportunity is to work within the CreativAI UC group to explore text diffusion models applied to symbolic music generation. In this internship, you will:
- Study diffusion models
- Implement and develop experiments to evaluate text diffusion models such as the recently released DiffusionGemma
- Attend weekly group meetings to stay up-to-date on topics within the Creative AI field https://creativai-uc.github.io/
(initially Tuesdays 12:30pm-1:30pm)

- Requirements:
* Advanced programming skills or equivalent
* Experience programming neural network models with PyTorch, Keras, TensorFlow, JAX, etc.

Semi-supervised automatic segmentation in cardiac magnetic resonance imaging

Medical image segmentation is a labor-intensive and time-consuming task, which presents a significant challenge in clinical practice. This is exacerbated by the scarcity of data manually annotated by experts, as the labeling of 3D volumes requires specialized personnel and extensive hours of work. In this context, artificial intelligence, and particularly semi-supervised learning, emerges as a promising solution: it allows for leveraging both a small set of labeled images and a larger volume of unlabeled images, reducing the reliance on manual annotations without sacrificing accuracy.
This work aims to utilize state-of-the-art neural network architectures, adapted for semi-supervised frameworks, to automatically segment various contrasts and 3D cardiac MRI maps, optimizing the annotation process and improving the model's accuracy even with limited labeled data.

Experimental Research in High-Speed Electronics (GaN Transistors)

I am looking for an undergraduate student to join a research project focused on the development and experimental validation of next-generation power converters.

The Technical Challenge:
The work involves supporting the experimental development of active gate driving techniques applied to Gallium Nitride (GaN) transistors in a half-bridge configuration.
The main tasks will be associated with the design, assembly, and validation of a high-bandwidth current measurement stage, including the selection and characterization of sensors or shunt resistors, parasitic effect analysis, calibration of the measurement chain, synchronization with voltage and gate signals, di/dt estimation, and switching energy calculation. The work builds upon progress already made by a highly experienced Master's student, who will provide laboratory supervision.

Desired Profile
Essential Requirements: Demonstrable practical experience in a laboratory, oscilloscope operation, component soldering, and PCB assembly/modification.

What you DON'T need: No prior knowledge of Power Electronics is required. We are looking for experimental rigor, an interest in instrumentation and high-speed signal analysis; you will learn the specific theory here.

Dynamic Analysis and Optimization Algorithms for the Control of Multivariable Systems

Many modern high-tech systems require controlling multiple critical variables simultaneously. Although this is traditionally addressed through cascade control schemes, current trends point towards Model Predictive Control (MPC) formulations, which consolidate regulation tasks into a single real-time optimization problem.

In this IPRE we will focus on analyzing the dynamic behavior of this type of formulations. The mathematical model of the system is already fully developed and available, so the student's work will consist of the theoretical characterization and simulation of the performance of the algorithms.

The project will address two main objectives:

System Dynamics Analysis: Study the properties of the resulting multivariable model, evaluating the couplings between states and control actions, as well as the linearization of the system.

Algorithm Evaluation (Solvers): Implement and compare different numerical optimization methods to evaluate the trade-off between mathematical precision, computational effort and feasibility of execution in real time.

Study application:
The practical case study where these tools will be evaluated corresponds to energy balance and current regulation in advanced power converters (used in smart grids, energy storage and electromobility).

Profile and Prerequisites:
Tools: MATLAB/Simulink or Python.

Prerequisites: Solid knowledge of Dynamic Systems and Automatic Control. Basic notions of mathematical optimization and programming.

Impact and Learning:
The student will develop advanced skills in the analysis of multivariable systems, formulation of optimization algorithms and simulation of predictive control strategies, skills highly transferable to both postgraduate research and technological development in the industry.

Development of new Astronomical Image Processing algorithms

Research the state of the art, looking for new research gaps or needs in the professional or amateur astronomical imaging communities and develop a new algorithm proposal.