Anomalous Decision Discovery using Inverse Reinforcement Learning

TL;DR

This paper introduces a novel IRL-based anomaly detection framework for autonomous vehicles that improves robustness and generalization in identifying unusual behaviors from perception data, outperforming existing methods.

Contribution

The paper proposes TRAP, a new IRL framework that enhances anomaly detection in AVs by addressing noise robustness and unseen scenario generalization through temporal reward learning.

Findings

Achieved 0.90 AUC and 82.2% F1-score on simulated trajectories.

Outperformed supervised and unsupervised baselines by 39% and 12% in recall and F1-score.

Demonstrated robustness to various noise types and unseen anomalies.

Abstract

Anomaly detection plays a critical role in Autonomous Vehicles (AVs) by identifying unusual behaviors through perception systems that could compromise safety and lead to hazardous situations. Current approaches, which often rely on predefined thresholds or supervised learning paradigms, exhibit reduced efficacy when confronted with unseen scenarios, sensor noise, and occlusions, leading to potential safety-critical failures. Moreover, supervised methods require large annotated datasets, limiting their real-world feasibility. To address these gaps, we propose an anomaly detection framework based on Inverse Reinforcement Learning (IRL) to infer latent driving intentions from sequential perception data, thus enabling robust identification. Specifically, we present Trajectory-Reward Guided Adaptive Pre-training (TRAP), a novel IRL framework for anomaly detection, to address two critical limitations of existing methods: noise robustness and generalization to unseen scenarios. Our core innovation is implicitly learning temporal credit assignments via reward and worst-case supervision. We leverage pre-training with variable-horizon sampling to maximize time-to-consequence, resulting in early detection of behavior deviation. Experiments on 14,000+ simulated trajectories demonstrate state-of-the-art performance, achieving 0.90 AUC and 82.2\% F1-score - outperforming similarly trained supervised and unsupervised baselines by 39\% on Recall and 12\% on F1-score, respectively. Similar performance is achieved while exhibiting robustness to various noise types and generalization to unseen anomaly types. Our code will be available at: https://github.com/abastola0/TRAP.git