Physical AI: Learned models for predicting the stability of object assemblies

A person glances at a stack of objects and has an intuition about whether it will hold. For a robot, answering that question exactly requires a computation whose cost explodes with the number of objects, since every way the assembly could come apart has to be considered. This project trains a learning-based model that quickly predicts how much force an arrangement can withstand before it collapses. What sets this project apart is that the learning is grounded in a physical model, which is what lets it generalize to scenes it has never seen.

Such a model is useful in many applications: palletizing and container loading, warehouse handling, safely transporting a pile of objects, avoiding moving too close to unstable structures. It also gives a manipulator the fast physical intuition it needs to select what object to pick such that other objects in contact don't loose stability.

You will have GPU workstations for training, and predictions will be checked against real assemblies disturbed by a robot arm, with a motion capture system measuring precisely when and how the stack gives way.

How to apply

To apply, send a complete application for the NSERC Canada Graduate Scholarship (CGS) Doctoral Program to Prof. Philippe Nadeau. Your application has to be sent through an institutional email address.

The CGS is the primary funding mechanism for PhD students in Canada, and competition for it is fierce. Preparing a full application up front may look like extra work, but it pays off: you'll arrive with a strong first draft of your CGS application well ahead of the deadline, improving your odds of being funded by the time you join the group.

Note: If you are considering applying for a Master's with thesis, the same steps apply but we will be using your application package to apply for the appropriate funding sources. Also, if you are an international student, your application will be submitted to the FRQ.