Abstract

While conventional reinforcement learning focuses on designing agents that can perform one task, meta-learning aims, instead, to solve the problem of designing agents that can generalize to different tasks (e.g., environments, obstacles, and goals) that were not considered during the design or the training of these agents. In this spirit, we consider the problem of training a provably safe Neural Network (NN) controller for uncertain nonlinear dynamical systems that can generalize to new tasks that were not present in the training data while preserving strong safety guarantees. I will present two complementary neurosymbolic approaches. In the first approach, I will show how to use ideas from symbolic control to provide guarantees on the training of NN controllers. In the second approach, I will show how to use NN to guide the design of symbolic controllers. I will discuss the theoretical guarantees governing the correctness and optimality of these neurosymbolic controllers and show experimental validation of our approach.