Sensing Anomalies as Potential Hazards: Datasets and Benchmarks

TL;DR

This paper introduces three new image datasets for detecting unusual visual patterns in robot environments, evaluates autoencoder-based anomaly detection methods, and aims to improve hazard detection for autonomous robots.

Contribution

The paper provides novel datasets with over 200,000 labeled frames for anomaly detection in robotic exploration scenarios and assesses autoencoder-based methods at multiple scales.

Findings

Autoencoder-based methods can identify anomalies in robot visual data.

The datasets cover diverse anomaly types and environments.

Evaluation results highlight the effectiveness of multi-scale autoencoders.

Abstract

We consider the problem of detecting, in the visual sensing data stream of an autonomous mobile robot, semantic patterns that are unusual (i.e., anomalous) with respect to the robot's previous experience in similar environments. These anomalies might indicate unforeseen hazards and, in scenarios where failure is costly, can be used to trigger an avoidance behavior. We contribute three novel image-based datasets acquired in robot exploration scenarios, comprising a total of more than 200k labeled frames, spanning various types of anomalies. On these datasets, we study the performance of an anomaly detection approach based on autoencoders operating at different scales.