About Me

I am a machine learning engineer at BMW AG working on autonomous driving.

Previously, I was a PhD student at Ulm University (Germany), supervised by Prof. Klaus Dietmayer and in collaboration with BMW AG. My research concentrated on 3D deep learning perception for autonomous driving applications, with a focus on robustness to adverse weather conditions. I did my undergraduate studies at the University of Siena (Italy), where I obtained a Master’s degree in Robotics and Automation Engineering.

In my spare time I enjoy travelling, running, bouldering, and reading a good book.

Publications

Label-Efficient Semantic Segmentation of LiDAR Point Clouds in Adverse Weather Conditions

Aldi Piroli, Vinzenz Dallabetta, Johannes Kopp, Marc Walessa, Daniel Meissner, Klaus Dietmayer

IEEE Robotics and Automation Letters (RA-L) 2024

Project Page
Video

SemanticSpray++: A Multimodal Dataset for Autonomous Driving in Wet Surface Conditions

Aldi Piroli, Vinzenz Dallabetta, Johannes Kopp, Marc Walessa, Daniel Meissner, Klaus Dietmayer

IEEE Intelligent Vehicles Symposium (IV 2024)

Project Page
Code

Towards Robust 3D Object Detection In Rainy Conditions

Aldi Piroli, Vinzenz Dallabetta, Johannes Kopp, Marc Walessa, Daniel Meissner, Klaus Dietmayer

IEEE International Conference on Intelligent Transportation Systems (ITSC) 2023

Project Page
Video

LS-VOS: Identifying Outliers in 3D Object Detections Using Latent Space Virtual Outlier Synthesis

Aldi Piroli, Vinzenz Dallabetta, Johannes Kopp, Marc Walessa, Daniel Meissner, Klaus Dietmayer

IEEE International Conference on Intelligent Transportation Systems (ITSC) 2023

Project Page

SemanticSpray Dataset

Aldi Piroli, Vinzenz Dallabetta, Johannes Kopp, Marc Walessa, Daniel Meissner, Klaus Dietmayer

Project Page
Video
Code

Energy-based Detection of Adverse Weather Effects in LiDAR Data

Aldi Piroli, Vinzenz Dallabetta, Johannes Kopp, Marc Walessa, Daniel Meissner, Klaus Dietmayer

IEEE Robotics and Automation Letters (RA-L) 2023

Project Page
Video

Detection of Condensed Vehicle Gas Exhaust in LiDAR Point Clouds

Aldi Piroli, Vinzenz Dallabetta, Marc Walessa, Daniel Meissner, Johannes Kopp, Klaus Dietmayer

IEEE International Conference on Intelligent Transportation Systems (ITSC) 2022

Robust 3D Object Detection in Cold Weather Conditions

Aldi Piroli, Vinzenz Dallabetta, Marc Walessa, Daniel Meissner, Johannes Kopp, Klaus Dietmayer

IEEE Intelligent Vehicles Symposium (IV) 2022

LMB Filter Based Tracking Allowing for Multiple Hypotheses in Object Reference Point Association

Martin Herrmann, Aldi Piroli, Jan Strohbeck, Johannes Müller, Michael Buchholz

IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI) 2020

Patents

Verfahren zum Trainieren eines Objektdetektors und damit ausgestattetes Kraftfahrzeug

Aldi Piroli, Vinzenz Dallabetta, Marc Walessa, Daniel Meissner

Verfahren und Detektionsvorrichtung zur lidarbasierten Objektdetektion bei nasser Fahrbahn und entsprechend eingerichtetes Kraftfahrzeug

Aldi Piroli, Vinzenz Dallabetta, Marc Walessa, Daniel Meissner

Verfahren und Datenverarbeitungseinrichtung zur lidarbasierten Umgebungserkennung und Kraftfahrzeug damit

Aldi Piroli, Vinzenz Dallabetta, Marc Walessa, Daniel Meissner

Verfahren zum Erzeugen von Trainingsdaten für ein lidarbasiertes Erkennen von Fremdfahrzeugen und zum Trainieren eines Detektionsmodells sowie Kraftfahrzeug

Aldi Piroli, Vinzenz Dallabetta, Marc Walessa, Daniel Meissner

Verfahren und Erkennungseinrichtung zum lidarbasierten Erkennen von Abgaswolken und Kraftfahrzeug

Aldi Piroli, Vinzenz Dallabetta, Marc Walessa, Daniel Meissner

Projects

A collection of for fun projects mostly on deep learning stuff.

sam_from_scratch | Code

From scratch implementation of the paper: “Segment Anything” (SAM) ICCV 2023

mae_from_scratch | Code

From scratch implementation of the paper: “Masked Autoencoders Are Scalable Vision Learners” CVPR 2022.

ViT_from_scratch | Code

Implementating Vision Transformers (ViT) from scatch, and deploying the model in C++.

mnist_cpp | Code

Training a network on the mnist_dataset in tensorflow to perform classification and digit detection and then deploying it in C++.

pointnet_from_scratch | Code

Implementing PointNet from scratch using pytorch.

flappy_bird_ai | Code

Flappy Bird game in Python. A genetic algorithm is used to create a data-set for an agent that is trained to always win the game.

poetry_classifier | Code

A model that can distinguish a line of poetry from two poets.

CV

2024-09 / Present : Machine Learning Engineer at BMW AG
2020-06 / 2024-09: PhD Student in 3D Deep Learning for Autonomous Driving
2017-09 / 2020-02: Master in Computer and Automation Engineering (110/110 cum laude)
2014-09 / 2017-07: Bachelor in Information Engineering