Iralab

05/04/2022

Congratulazioni a tutti i tesisti!

11/11/2021

Un anno di riparazioni e tante sere spese al cart

23/01/2020

Small steps for our

22/01/2020

Great news: our paper "Global visual localization in LiDAR-maps through shared 2D-3D embedding space" has been accepted for the conference in Paris.

You can download our preprint on arXiv:
https://arxiv.org/abs/1910.04871

Abstract: Global localization is an important and widely studied problem for many robotic applications. Place recognition approaches can be exploited to solve this task, e.g., in the autonomous driving field. While most vision-based approaches match an image w.r.t an image database, global visual localization within LiDAR-maps remains fairly unexplored, even though the path toward high definition 3D maps, produced mainly from LiDARs, is clear. In this work we leverage DNN approaches to create a shared embedding space between images and LiDAR-maps, allowing for image to 3D-LiDAR place recognition. We trained a 2D and a 3D Deep Neural Networks (DNNs) that create embeddings, respectively from images and from point clouds, that are close to each other whether they refer to the same place. An extensive experimental activity is presented to assess the effectiveness of the approach w.r.t. different learning methods, network architectures, and loss functions. All the evaluations have been performed using the Oxford Robotcar Dataset, which encompasses a wide range of weather and light conditions.

While most vision-based approaches match an image with respect to an image database, global visual localization within LiDAR-maps remains fairly unexplored. ...

17/12/2019

Nice to be all together again 😊 Iralab Christmas Dinner🎄

17/12/2019

Nice to be all together again 😊IraLab Christmas dinner!🎄

13/11/2019

When the dream came true..

28/10/2019

Excellent work Daniele! Congratulations!

24/10/2019

Congratulations to and for their graduation!!!

24/10/2019

are you ready?

01/10/2019

Celebrating our project management tool! Today we’ve reached 1K of documented activities! From 2012/13 we’ve tracked thousands work hours, issues, and valuable work of more than 70 students and collaborators, creating a huge knowledge base for future research. We would like to thank all who has been involved in this great work!

24/09/2019

submission: Global visual localization in LiDAR-maps through shared 2D-3D embedding space

While most vision-based approaches match an image with respect to an image database, global visual localization within LiDAR-maps remains fairly unexplored. In this work we leverage DNN approaches to create a shared embedding space between images and LiDAR-maps, allowing for image to 3D-LiDAR place recognition.