ALFA (AirLab Failure and Anomaly Dataset)

ALFA (AirLab Failure and Anomaly Dataset) from CMU's AirLab is a real-world flight dataset for fault detection and isolation (FDI) and anomaly detection (AD) on a fixed-wing UAV. It contains processed data for 47 autonomous flights covering eight fault types: 23 sudden full engine (power loss) failures and 24 scenarios of seven control-surface (actuator) faults including elevator stuck at zero, aileron stuck (left/right/both), rudder stuck (left/right/zero), and combined rudder & aileron faults, plus normal (no-fault) baseline flights. Across the processed sequences there is about 66 minutes of normal flight and 13 minutes of post-fault flight; the full archive additionally includes many hours of raw autonomous, autopilot-assisted, and manual flight data with tens of fault scenarios. The aircraft is a custom-modified Carbon-Z T-28 model plane (2 m wingspan, single electric engine, ailerons/flaperons/elevator/rudder) equipped with a Pixhawk autopilot (modified ArduPilot 3.9.0beta1, MAVLink 2.0), an NVIDIA Jetson TX2 onboard computer, a pitot tube, and a GPS module. Recorded telemetry (typically 20-25 Hz, with some 4-5 Hz topics and ~5 Hz ground-truth fault-status topics) includes commanded and measured roll/pitch/yaw, airspeed and velocity, IMU state and accelerations, GPS/global and local position, wind estimation, path/altitude/airspeed tracking errors, and setpoint commands, along with ground-truth labels for engine, aileron, rudder, and elevator faults. Data is distributed as ROS .bag files plus processed .csv and .mat (MATLAB) exports, plus raw bag files, TX2 telemetry logs, and Pixhawk dataflash logs; cross-platform reader/filter/iterator and evaluation tools (C++, Python, MATLAB) and custom ROS message definitions are provided at github.com/castacks/alfa-dataset-tools (BSD-3 licensed). The dataset is hosted publicly on CMU KiltHub under a CC BY 4.0 license (~1.7 GB compressed / ~12.5 GB uncompressed across processed/raw/dataflash/telemetry archives). Authored by Azarakhsh Keipour, Mohammadreza Mousaei, and Sebastian Scherer (IJRR 2021; arXiv:1907.06268). No camera/vision data — it is telemetry/sensor logs only.

ALFA (AirLab Failure and Anomaly Dataset) from CMU's AirLab is a real-world flight dataset for fault detection and isolation (FDI) and anomaly detection (AD) on a fixed-wing UAV. It contains processed data for 47 autonomous flights covering eight fault types: 23 sudden full engine (power loss) failures and 24 scenarios of seven control-surface (actuator) faults including elevator stuck at zero, aileron stuck (left/right/both), rudder stuck (left/right/zero), and combined rudder & aileron faults, plus normal (no-fault) baseline flights. Across the processed sequences there is about 66 minutes of normal flight and 13 minutes of post-fault flight; the full archive additionally includes many hours of raw autonomous, autopilot-assisted, and manual flight data with tens of fault scenarios. The aircraft is a custom-modified Carbon-Z T-28 model plane (2 m wingspan, single electric engine, ailerons/flaperons/elevator/rudder) equipped with a Pixhawk autopilot (modified ArduPilot 3.9.0beta1, MAVLink 2.0), an NVIDIA Jetson TX2 onboard computer, a pitot tube, and a GPS module. Recorded telemetry (typically 20-25 Hz, with some 4-5 Hz topics and ~5 Hz ground-truth fault-status topics) includes commanded and measured roll/pitch/yaw, airspeed and velocity, IMU state and accelerations, GPS/global and local position, wind estimation, path/altitude/airspeed tracking errors, and setpoint commands, along with ground-truth labels for engine, aileron, rudder, and elevator faults. Data is distributed as ROS .bag files plus processed .csv and .mat (MATLAB) exports, plus raw bag files, TX2 telemetry logs, and Pixhawk dataflash logs; cross-platform reader/filter/iterator and evaluation tools (C++, Python, MATLAB) and custom ROS message definitions are provided at github.com/castacks/alfa-dataset-tools (BSD-3 licensed). The dataset is hosted publicly on CMU KiltHub under a CC BY 4.0 license (~1.7 GB compressed / ~12.5 GB uncompressed across processed/raw/dataflash/telemetry archives). Authored by Azarakhsh Keipour, Mohammadreza Mousaei, and Sebastian Scherer (IJRR 2021; arXiv:1907.06268). No camera/vision data — it is telemetry/sensor logs only.

has-success-labels · ALFA (AirLab Failure and Anomaly Dataset) from CMU's AirLab is a real-world flight dataset for fault detection and isolation (FDI) and anomaly detection (AD) on a fixed-wing UAV. It contains processed data for 47 autonomous flights covering eight fault types: 23 sudden full engine (power loss) failures and 24 scenarios of seven control-surface (actuator) faults including elevator stuck at zero, aileron stuck (left/right/both), rudder stuck (left/right/zero), and combined rudder & aileron faults, plus normal (no-fault) baseline flights. Across the processed sequences there is about 66 minutes of normal flight and 13 minutes of post-fault flight; the full archive additionally includes many hours of raw autonomous, autopilot-assisted, and manual flight data with tens of fault scenarios. The aircraft is a custom-modified Carbon-Z T-28 model plane (2 m wingspan, single electric engine, ailerons/flaperons/elevator/rudder) equipped with a Pixhawk autopilot (modified ArduPilot 3.9.0beta1, MAVLink 2.0), an NVIDIA Jetson TX2 onboard computer, a pitot tube, and a GPS module. Recorded telemetry (typically 20-25 Hz, with some 4-5 Hz topics and ~5 Hz ground-truth fault-status topics) includes commanded and measured roll/pitch/yaw, airspeed and velocity, IMU state and accelerations, GPS/global and local position, wind estimation, path/altitude/airspeed tracking errors, and setpoint commands, along with ground-truth labels for engine, aileron, rudder, and elevator faults. Data is distributed as ROS .bag files plus processed .csv and .mat (MATLAB) exports, plus raw bag files, TX2 telemetry logs, and Pixhawk dataflash logs; cross-platform reader/filter/iterator and evaluation tools (C++, Python, MATLAB) and custom ROS message definitions are provided at github.com/castacks/alfa-dataset-tools (BSD-3 licensed). The dataset is hosted publicly on CMU KiltHub under a CC BY 4.0 license (~1.7 GB compressed / ~12.5 GB uncompressed across processed/raw/dataflash/telemetry archives). Authored by Azarakhsh Keipour, Mohammadreza Mousaei, and Sebastian Scherer (IJRR 2021; arXiv:1907.06268). No camera/vision data — it is telemetry/sensor logs only.

ALFA (AirLab Failure and Anomaly Dataset) from CMU's AirLab is a real-world flight dataset for fault detection and isolation (FDI) and anomaly detection (AD) on a fixed-wing UAV. It contains processed data for 47 autonomous flights covering eight fault types: 23 sudden full engine (power loss) failures and 24 scenarios of seven control-surface (actuator) faults including elevator stuck at zero, aileron stuck (left/right/both), rudder stuck (left/right/zero), and combined rudder & aileron faults, plus normal (no-fault) baseline flights. Across the processed sequences there is about 66 minutes of normal flight and 13 minutes of post-fault flight; the full archive additionally includes many hours of raw autonomous, autopilot-assisted, and manual flight data with tens of fault scenarios. The aircraft is a custom-modified Carbon-Z T-28 model plane (2 m wingspan, single electric engine, ailerons/flaperons/elevator/rudder) equipped with a Pixhawk autopilot (modified ArduPilot 3.9.0beta1, MAVLink 2.0), an NVIDIA Jetson TX2 onboard computer, a pitot tube, and a GPS module. Recorded telemetry (typically 20-25 Hz, with some 4-5 Hz topics and ~5 Hz ground-truth fault-status topics) includes commanded and measured roll/pitch/yaw, airspeed and velocity, IMU state and accelerations, GPS/global and local position, wind estimation, path/altitude/airspeed tracking errors, and setpoint commands, along with ground-truth labels for engine, aileron, rudder, and elevator faults. Data is distributed as ROS .bag files plus processed .csv and .mat (MATLAB) exports, plus raw bag files, TX2 telemetry logs, and Pixhawk dataflash logs; cross-platform reader/filter/iterator and evaluation tools (C++, Python, MATLAB) and custom ROS message definitions are provided at github.com/castacks/alfa-dataset-tools (BSD-3 licensed). The dataset is hosted publicly on CMU KiltHub under a CC BY 4.0 license (~1.7 GB compressed / ~12.5 GB uncompressed across processed/raw/dataflash/telemetry archives). Authored by Azarakhsh Keipour, Mohammadreza Mousaei, and Sebastian Scherer (IJRR 2021; arXiv:1907.06268). No camera/vision data — it is telemetry/sensor logs only.

Open · CC-BY-4.0 · custom · rosbag