GbO — Guaranteed by Optimization

This project’s goal is to advance the state of the art of optimization-based design in robotics.

The researchers will develop and study an original method to automatically generate control software that guarantees a number of dynamic properties such as stability, constraint satisfaction, adaptability, and resilience. This method will enable a designer to inject into the optimization process their previous knowledge of the way in which the desired properties can be guaranteed/checked. For its research, the team will draw from automatic design, control theory, and verification processes.

The researchers will apply their ideas to various types of robots, including land-based, aquatic, and flying robots, as well as robotic arms.

They will look at different classes of robotic systems, each of which poses specific design challenges: single robots, multi-robot systems (teams of robots whose interactions can be defined at the design stage), and robot swarms (large groups of robots whose interactions are defined only at runtime).

They will verify their theoretical results by conducting experiments with simulations, then with actual robots.

Coordination : Mauro Birratari, IRIDIA, Ecole Polytechnique de Bruxelles
Partner : Emmanuele Garone, SAAS

Created on September 3, 2020