Elastic ODYN: Differentiable Optimization for Infeasible Control and Learning in Robotics
The episode discusses ELASTIC ODYN, a primal-dual non-interior-point QP solver that handles infeasibility in robotics using smooth squared two elastic relaxations. The hosts explore how this framework provides stability for real-time control and learning by maintaining differentiable layers even when constraints are violated. They conclude that the work offers a robust method for building resilient AI systems capable of handling messy physical interactions.