LSCOE · SYSTEC · FEUP · U.Porto

Small Scale
Autonomous Racing
UP Project

High-performance motion planning, optimal control, and perception research for autonomous racing vehicles — bridging theory and practice at the University of Porto.

View Research Our Platform
9+
MSc Theses
5+
Internships
2
Research Labs
∞
Iterations / Day

About the Project

Where control theory meets the race track

The UP2Speed activities at LSCOE · SYSTEC · FEUP investigate high-performance control, motion planning, perception and learning for autonomous racing vehicles at small scale.

The project is tightly integrated with MSc theses and curricular internships in Electrical and Computers Engineering, providing students with hands-on experience in modern control, robotics and autonomous driving.

Optimal Control Motion Planning Perception Reinforcement Learning MPC Computer Vision
UP2SPEED autonomous racing vehicles
FEUP
Porto, Portugal

Research Framework

Three Pillars

Every UP2Speed activity is grounded in three interconnected research pillars. Each pillar poses distinct scientific questions and requires both theoretical depth and experimental validation.

Perception
Detect, localize and understand the environment in real time. Covers camera-based track and obstacle recognition, state estimation from onboard and external sensors, sensor fusion, and SLAM for reliable localization on the race track.
Planning
Generate fast, feasible and safe trajectories. Covers minimum-lap-time trajectory optimization, online replanning, collision avoidance and decision-making under constraints and dynamic environments.
Control
Track the planned trajectory precisely and at speed. Covers model predictive control, learning-based MPC, robust and adaptive control, real-time embedded implementation and closed-loop stability under model mismatch and actuation limits.

Open Research Questions

Challenges

Autonomous racing at small scale still poses a rich set of open problems. Each challenge is tagged by pillar and by orientation: Software for algorithm and control problems, Hardware for systems and integration problems.

Perception
Software
Robust visual detection
Keeping track and obstacle detection reliable under varying lighting, partial occlusion, and motion blur. Requires robust vision pipelines and adaptive algorithms.
Perception Control
Software · Hardware
State estimation with limited sensors
Fusing camera data with motion capture or IMU into a stable, low-latency state estimate for real-time control. Balancing accuracy, delay, and computational cost.
Planning
Software
Real-time trajectory optimization
Computing fast, constraint-satisfying trajectories online. Requires efficient solvers, warm-starting strategies, and handling of dynamic obstacles or changing conditions.
Control Planning
Software · Hardware
Embedded real-time execution
Deploying control and planning algorithms on onboard computers with limited resources. Meeting timing requirements, reducing latency, and ensuring reliability during live experiments.
Planning
Software
Safe decision-making at speed
Handling overtaking, dynamic obstacle avoidance and lane changes without sacrificing safety or performance. Requires planning with safety guarantees and uncertainty awareness.
Control
Software
Model learning and adaptation
Learning and correcting vehicle model errors from data to improve MPC predictions. Includes Gaussian process regression, iterative learning and adaptive control for changing conditions.
Control
Software
High-speed tracking under uncertainty
Minimising tracking error while operating near the vehicle's physical limits. Robust MPC, tube-based approaches and disturbance rejection are key tools to explore.
Planning Control
Hardware · Software
Multi-vehicle communication and coordination
Reliable low-latency communication between vehicles and infrastructure. Enabling cooperative scenarios, shared perception and coordinated racing with multiple cars on track.

Hardware & Infrastructure

Experimental Platform

The experimental setup combines a race track, motion capture and onboard sensing for real-time control and perception, inspired by established autonomous RC car racing testbeds.

Reduced-Scale Vehicles
Configurable RC race cars with embedded computation enable fast algorithm iteration while remaining safe and cost-effective.
Motion Capture & Sensing
Onboard cameras and external vision systems provide rich state estimation for real-time control and perception research.
Embedded Computation
Communication infrastructure and onboard embedded systems bridge perception, planning, and control end-to-end.
Research Scenarios
01
Time-Trial Racing
Minimum-time optimal trajectories
02
Double Lane Change
Evasive manoeuvre control
03
Multi-Vehicle Interaction
Cooperative driving scenarios

Academic Work

Research & Theses

Master Theses

Mário Miguel Moniz Xavier
Motion Planning and Control for a Vehicle Performing a Double Lane Change Maneuver
ongoing M.EEC · FEUP Supervisor: Luís Tiago Paiva and Thien Nguyen
Repository
Pedro Alexandre Mendes Lopes
Formal Methods for Collision Avoidance of Autonomous Racing Vehicles
ongoing M.EEC · FEUP Supervisor: Luís Tiago Paiva
Repository
Vasco Machado Perdigão
Perception System for Monitoring and Control of Several Autonomous Racing Vehicles and Environment
ongoing M.EEC · FEUP Supervisor: Luís Tiago Paiva
Repository
Raúl Miguel Domingues Moreira da Silva
Control of a Self-Balancing Motorcycle
2026 M.EEC · FEUP Supervisor: Luís Tiago Paiva
Repository 16 / 20
Martim Luís Pedreira Iglésias
Control of a Four-In-Wheel Motor Electric Vehicle For Autonomous Racing
2025 M.EEC · FEUP Supervisor: Luís Tiago Paiva
Repository 17 / 20
Tomás Gomes Monteiro Sanhudo
Collision Avoidance Strategies for Autonomous Racing Vehicles
2025 M.EEC · FEUP Supervisor: Luís Tiago Paiva
Repository 18 / 20
Duarte Lima Silva
A Digital Twin Framework for Autonomous Racing Vehicles
2025 M.EEC · FEUP Supervisors: Luís Tiago Paiva & Fernando Fontes
Repository 16 / 20
Gonçalo Faria Ferreira Pimentel da Mota
Autonomous Vehicle Racing - Perception
2024 M.EEC · FEUP Supervisors: Fernando Fontes & Paulo Lopes dos Santos
Repository 16 / 20
João Paulo de Jesus Marques Sampaio
Autonomous Vehicle Racing - Learning Control
2024 M.EEC · FEUP Supervisors: Fernando Fontes & Mª Rosário de Pinho
Repository 16 / 20

Internships

Pedro Simão Valente Sousa
Detection and Control of a Small-Scale RC Vehicle using Computer Vision
ongoing L.EEC · FEUP · ISR Supervisors: Luís Tiago Paiva and Conrado Costa
Diogo Barbosa Navio
Environment and Obstacle Recognition with Computer Vision
ongoing L.EEC · FEUP · ISR Supervisors: Luís Tiago Paiva and Conrado Costa
Miguel Marques Alves
Perception System via Computer Vision
2025 L.EEC · FEUP · ISR Supervisors: Luís Tiago Paiva and Conrado Costa
Beatriz Bonifácio Martins
Computer Vision for Monitoring and Control of Small-Scale Vehicles (~1:43)
2025 L.EEC · FEUP · ISR Supervisors: Luís Tiago Paiva and Conrado Costa
Gabriel Sousa Oliveira
Introduction to Trajectory-Following Control Systems
2025 L.EEC · FEUP · ISR Supervisors: Luís Tiago Paiva and Conrado Costa

People

Research Team

Luís Tiago Paiva
Project Coordinator
LSCOE · SYSTEC · FEUP
University of Porto
Get to Know
Fernando Fontes
Lab Coordinator
LSCOE · SYSTEC · FEUP
University of Porto
Get to Know
Thien Nguyen
Post Doc
LSCOE · SYSTEC · FEUP
University of Porto
Get to Know
Conrado Costa
PhD Student
LSCOE · SYSTEC · FEUP
University of Porto
Get to Know

Students contribute through MSc theses, curricular internships and project‑based courses in collaboration with ISR Porto, covering control, planning, perception, embedded systems and software development for autonomous vehicles.

Get in Touch

Contact Us

Address
Laboratory of Control and Systems Optimization (LSCOE)
SYSTEC – Research Center for Systems and Technologies
Faculty of Engineering, University of Porto (FEUP)
Rua Dr. Roberto Frias, s/n · 4200‑465 Porto, Portugal
Research Group
LSCOE · SYSTEC · FEUP
In collaboration with ISR Porto