Ferran Alet

PhD student at Massachusetts Institute of Technology

Email: {mylastname}@mit.edu

CV / Linkedin

About me

I am PhD student at the Learning and Intelligent Systems Lab at the Massachusetts Institute of Technology with Leslie Kaelbling and Tomas Lozano-Perez. My interests are in the intersection of Machine Learning and Robotics. I am currently being fully funded by La Caixa fellowship.

Current Projects

Robotic Manipulation in cluttered environments

During the last two years I have been responsible for the High Level Planner of the MIT-Princeton team in the Amazon Robotics Challenge Competition.

In 2016 we placed 3rd in the Stowing Task and we won the Stowing Task in 2017. You can find more information on the setting of 2017 and the vision part of our robot in this paper.

Finding Frequent Entities in Continuous Data

Accepted at IJCAI '18, to be presented in Stockholm this July! We recently proposed an online algorithm to detect important entities in a data stream. For instance, a robot may see tens or hundreds of people in a single day, yet only care about a small set of people it regularly interacts with. While conventional clustering algorithms (such as k-means, DBSCAN) are overwhelmed by the vast proportion of uninteresting data, we propose an algorithm with provable guarantees that is able to identify this small set of identities in an online way.

You can find all the details here.


This last summer I did an internship at Google Research Europe, where I proposed the first working prototype to automatically create ads from raw YouTube content, in a totally unsupervised way.

In 2015 I did another internship in Google Zurich. There, I did research in Deep Learning to improve the handwriting recognizer used, for instance, in Google Translate.

Past projects and other interests

Self-driving RC car

With a partner at CFIS, we hacked the control of a cheap RC car with Arduino. From the images captured from a mobile phone (in her hands in the video) the car was able to drive itself through obstacles (red paper) better than we could!

This may seem implausible given the low speed of the self-driving car; however, since the original RC car was too fast and couldn't move in a straight line, we repeatedly went out of track when we tried to go faster than the self-driving car. The self-driving car manages to move smoothly by repeatedly pushing and stopping the acceleration button at high frequency.

Mathematics and Physics

I studied two undergraduate degrees: Mathematics and Engineering Physics. I was able to do it thanks to CFIS, a scholarship center in Barcelona that gives scholarships to the most promising students in Spain to do two degrees simultaneously. In Mathematics, I am passionate about Combinatorics, Graph Theory, Number Theory, Probability and Statistics. In Physics, I am interested in Mechanics, Statistical Physics, Network Science and Quantum Physics.

Talk as the winner of the Poincare research thesis award

Programming competitions

I love doing Programming competitions, where participants are asked to write codes that solve several algorithmic problems. Solving them, involves a combination of imagination, logical thinking, algorithmic and coding abilities and great teamwork.

During my undergrad I competed with several teammates from UPC in the ACM-ICPC South-Western European Regional, becoming the most decorated participant in the history of the competition (20 years, 6 countries). You can read an example of problem set and solutions of this competition.

Bootstrap Thumbnail Third