Carlos Mastalli

Elastic ODYN can optimize robotic control tasks even when faced with conflicting objectives and infeasible constraints, outperforming traditional solvers in robustness and convergence.

Achieve faster, more accurate robot system identification by embedding physical constraints directly into a Bayesian estimation framework, leading to measurable improvements in model predictive control tracking performance.

Forget interior-point methods: ODYN tackles tough quadratic programs in robotics and AI with a novel all-shifted approach, showing superior warm-starting and convergence without needing linearly independent constraints.