Datasets:

aykutkabaoglu
/

uav-flight-anomaly-dataset

@@ -15,7 +15,7 @@ tags:
 ---
-# Dataset Card for UAV-SEAD
 ## Citation
@@ -29,7 +29,7 @@ tags:
 }
 ```
-## Dataset Details
 ### Dataset Description
 UAV-SEAD (State Estimation Anomaly Dataset) is a comprehensive, large-scale, real-world flight dataset designed to advance research in Fault Detection and Identification (FDI) for Unmanned Aerial Vehicles. It contains 1,396 categorized real-world flight logs totaling over 52.4 hours of flight data. Unlike existing benchmarks, it avoids synthetic fault injection, providing naturally occurring anomalies across diverse environmental and hardware configurations.
@@ -42,46 +42,34 @@ UAV-SEAD (State Estimation Anomaly Dataset) is a comprehensive, large-scale, rea
 ### Toolbox for Dataset Preprocessing and Annotation
 - https://github.com/aykutkabaoglu/ulog_annotation_tool
-## Uses
-### Direct Use
-- Benchmarking multivariate time-series anomaly detection algorithms.
-- Validating the robustness of Extended Kalman Filter (EKF) and other state estimators.
-- Developing predictive maintenance and fault identification systems for quadrotors.
-## Dataset Structure
 The dataset is structured as a collection of multivariate time-series sequences.
 - **Point Anomalies:** Spikes and outliers in sensor readings.
 - **Contextual Anomalies:** Readings that are only anomalous relative to the flight state (e.g., zero altitude during thrust).
 - **Collective Anomalies:** Sustained drifts or patterns indicating gradual system failure.
-## Dataset Creation
-### Curation Rationale
-Research in UAV safety is currently limited by a lack of real-world data. Most datasets are simulated or contain artificial, "clean" faults. UAV-SEAD was created to capture the messy, stochastic nature of real-world hardware failures and environmental noise.
-### Source Data
 #### Data Collection and Processing
 Data was harvested from a fleet of custom-built, PX4-based quadrotors.
-- **Hardware:** Multiple airframes with variations in motor sizes, weights, and sensor suites.
 - **Environments:** Flight missions conducted in indoor hangars, warehouses, and open outdoor spaces.
-- **Cleaning:** Raw ULog data was converted to CSV, synchronized to a common time-base, and validated for continuity.
-### Annotations
 #### Annotation process
 Annotation was performed via **Physically-Constrained Expert Evaluation**. Human experts identified state estimation divergences by comparing autopilot estimates with real-world observed outcomes. Labels were verified against physical flight dynamics to ensure the statistical anomalies correlate with actual physical events.
-#### Who are the annotators?
-Domain experts in UAV flight control and autonomous systems.
 ### Bias, Risks, and Limitations of Data Collection
 - **Platform Specificity:** Limited to multi-rotor dynamics (quadrotors). Limited outdoor flights with GPS availability.
-## Dataset Statistics
 ### Annotated Data Flight Source Statistics
 Annotated Number of Flight: 1396
@@ -200,7 +188,6 @@ The dataset follows a **Superset Schema** of 1897 unique columns. Fields are gro
 ###  Dataset Features
 Major PX4 topics that can mostly be found in flight logs and their descriptions:

 ---
+# DATASET CARD for UAV-SEAD
 ## Citation
 }
 ```
+## DATASET INFORMATION
 ### Dataset Description
 UAV-SEAD (State Estimation Anomaly Dataset) is a comprehensive, large-scale, real-world flight dataset designed to advance research in Fault Detection and Identification (FDI) for Unmanned Aerial Vehicles. It contains 1,396 categorized real-world flight logs totaling over 52.4 hours of flight data. Unlike existing benchmarks, it avoids synthetic fault injection, providing naturally occurring anomalies across diverse environmental and hardware configurations.
 ### Toolbox for Dataset Preprocessing and Annotation
 - https://github.com/aykutkabaoglu/ulog_annotation_tool
+- Use readme in the `ulog_annotation_tool` repository to start using dataset. The first step is converting raw flight logs into csv files. Then the flight logs can be used for plotting and labeling.
+## DATASET CREATION AND USAGE
+### Dataset Structure
 The dataset is structured as a collection of multivariate time-series sequences.
 - **Point Anomalies:** Spikes and outliers in sensor readings.
 - **Contextual Anomalies:** Readings that are only anomalous relative to the flight state (e.g., zero altitude during thrust).
 - **Collective Anomalies:** Sustained drifts or patterns indicating gradual system failure.
+### Use Cases
+- Benchmarking multivariate time-series anomaly detection algorithms.
+- Developing predictive maintenance and fault identification systems for quadrotors.
 #### Data Collection and Processing
 Data was harvested from a fleet of custom-built, PX4-based quadrotors.
+- **Hardware:** Multiple airframes with variations in motor sizes, weights, and sensor suites. Check paper for hardware details.
 - **Environments:** Flight missions conducted in indoor hangars, warehouses, and open outdoor spaces.
 #### Annotation process
 Annotation was performed via **Physically-Constrained Expert Evaluation**. Human experts identified state estimation divergences by comparing autopilot estimates with real-world observed outcomes. Labels were verified against physical flight dynamics to ensure the statistical anomalies correlate with actual physical events.
 ### Bias, Risks, and Limitations of Data Collection
 - **Platform Specificity:** Limited to multi-rotor dynamics (quadrotors). Limited outdoor flights with GPS availability.
+## DATASET STATISTICS
 ### Annotated Data Flight Source Statistics
 Annotated Number of Flight: 1396
 ###  Dataset Features
 Major PX4 topics that can mostly be found in flight logs and their descriptions: