Learning Certifiably Robust Controllers Using Fragile Perception

TL;DR

This paper presents a method to synthesize certifiably robust controllers that can handle perception errors in robotic systems by combining Gaussian process-based state estimation with robust neural network controllers, validated through simulations.

Contribution

It introduces a novel approach that integrates Gaussian process state estimation with neural network control to ensure robustness against perception errors.

Findings

Successfully synthesizes controllers that handle perception uncertainties.

Demonstrates effectiveness in a vision-based lane-keeping simulation.

Provides a framework for certifiable robustness in perception-driven control.

Abstract

Advances in computer vision and machine learning enable robots to perceive their surroundings in powerful new ways, but these perception modules have well-known fragilities. We consider the problem of synthesizing a safe controller that is robust despite perception errors. The proposed method constructs a state estimator based on Gaussian processes with input-dependent noises. This estimator computes a high-confidence set for the actual state given a perceived state. Then, a robust neural network controller is synthesized that can provably handle the state uncertainty. Furthermore, an adaptive sampling algorithm is proposed to jointly improve the estimator and controller. Simulation experiments, including a realistic vision-based lane-keeping example in CARLA, illustrate the promise of the proposed approach in synthesizing robust controllers with deep-learning-based perception.