Detecting and Mitigating System-Level Anomalies of Vision-Based Controllers

TL;DR

This paper presents a run-time anomaly detection and mitigation framework for vision-based controllers in autonomous systems, improving safety by identifying and handling system-level failures caused by out-of-distribution inputs.

Contribution

It introduces a reachability-based offline stress-testing method combined with online classification and a fallback controller to enhance system safety.

Findings

Effective detection of system-level anomalies in autonomous aircraft taxiing.

Outperforms prediction error-based detection and ensembling methods.

Improves overall safety and robustness of vision-based autonomous systems.

Abstract

Autonomous systems, such as self-driving cars and drones, have made significant strides in recent years by leveraging visual inputs and machine learning for decision-making and control. Despite their impressive performance, these vision-based controllers can make erroneous predictions when faced with novel or out-of-distribution inputs. Such errors can cascade to catastrophic system failures and compromise system safety. In this work, we introduce a run-time anomaly monitor to detect and mitigate such closed-loop, system-level failures. Specifically, we leverage a reachability-based framework to stress-test the vision-based controller offline and mine its system-level failures. This data is then used to train a classifier that is leveraged online to flag inputs that might cause system breakdowns. The anomaly detector highlights issues that transcend individual modules and pertain to the safety of the overall system. We also design a fallback controller that robustly handles these detected anomalies to preserve system safety. We validate the proposed approach on an autonomous aircraft taxiing system that uses a vision-based controller for taxiing. Our results show the efficacy of the proposed approach in identifying and handling system-level anomalies, outperforming methods such as prediction error-based detection, and ensembling, thereby enhancing the overall safety and robustness of autonomous systems.