GRAD: Real-Time Gated Recurrent Anomaly Detection in Autonomous Vehicle Sensors Using Reinforced EMA and Multi-Stage Sliding Window Techniques

TL;DR

GRAD addresses real-time anomaly detection in autonomous vehicle sensor data by integrating a reinforced EMA (REMA) with adaptive thresholds, a multi-stage sliding window (MS-SW) for multi-scale feature extraction, and a lightweight GRU classifier, followed by a recovery module to rectify damaged readings. The approach yields high anomaly-detection and -classification performance while keeping computational costs low for onboard deployment, as demonstrated on MMITSS and Zurich GPS datasets with injected anomalies. REMA provides fast, robust outlier bounds, while GRU-based detection leverages MS-SW features to capture both short- and long-term patterns, achieving an average anomaly F1 of 97.6% and normal F1 of 99.4% across datasets. The work offers a practical, real-time anomaly-detection framework for autonomous vehicles with strong potential for safe operation and efficient resource use.

Abstract

This paper introduces GRAD, a real-time anomaly detection method for autonomous vehicle sensors that integrates statistical analysis and deep learning to ensure the reliability of sensor data. The proposed approach combines the Reinforced Exponential Moving Average (REMA), which adapts smoothing factors and thresholding for outlier detection, with the Multi-Stage Sliding Window (MS-SW) technique for capturing both short- and long-term patterns. These features are processed using a lightweight Gated Recurrent Unit (GRU) model, which detects and classifies anomalies based on bias types, while a recovery module restores damaged sensor data to ensure continuous system operation. GRAD has a lightweight architecture consisting of two layers of GRU with a limited number of neurons that make it appropriate for real-time applications while maintaining high detection accuracy. The GRAD framework achieved remarkable performance in anomaly detection and classification. The model demonstrated an overall F1-score of 97.6% for abnormal data and 99.4% for normal data, signifying its high accuracy in distinguishing between normal and anomalous sensor data. Regarding the anomaly classification, GRAD successfully categorized different anomaly types with high precision, enabling the recovery module to accurately restore damaged sensor data. Relative to analogous studies, GRAD surpasses current models by attaining a balance between elevated detection accuracy and diminished computational expense. These results demonstrate GRAD's potential as a reliable and efficient solution for real-time anomaly detection in autonomous vehicle systems, guaranteeing safe vehicle operation with minimal computational overhead.

Figures (9)

• Figure 1: Illustration of the Waymo autonomous vehicle with hypothetical positions of its main sensors. The actual sensor placements may vary in real-world.
• Figure 2: Types of GPS anomalies in location data of the vehicles.
• Figure 3: GRU structure highlighting its gates and operation.
• Figure 4: Illustrating REMA bounds and value adjustments.
• Figure 5: GRAD Framework using REMA and GRU for anomaly detection, with a recovery module to ensure system reliability.