• autonomous driving
• autonomous racing
• FastSLAM
• Formula SAE
• mobile robotics
• SLAM

Mapping an unknown environment exploiting sensors measurements and simultaneously localize a mobile robotic platform within it represents a crucial task for
modern robotics.
This problem is typically referred to as Simultaneous Localization And Mapping
(SLAM) and solving it efficiently is still one of the most challenging task of mobile robotics.
This work is focused on the SLAM problem in the autonomous racing context, where
precision, reliability and performance need to be taken to an extreme level.
A series of competitions, coordinated by the Society of Automotive Engineers (SAE),
challenge every year teams from universities all over the world to compete with small,
formula-style autonomous vehicles.
A survey about the state-of-the-art SLAM solutions in a racing context, presented
in this master thesis, leads to a final implementation which wants to be reliable and
accurate in the result.
Autonomous race car softwares also needs to respect important constraints in terms
of computational performance. Different techniques and optimizations has been
applied in order to reduce the computational time and to respect the real-time con-
straint.
The implemented solution has been firstly validated in a simulated environment, enabling in-depth evaluation of the final algorithm.
Subsequently, the algorithm was deployed on the real autonomous vehicle, called
Kerub-EVO, of the University of Pisa Formula SAE team for on-track test ses-
sions.