This dataset contains wind condition and flight process data totaling over 46 hours from 270 flights involving three unmanned aerial vehicles (UAVs). The dataset covers flight data under various scenario settings, including constant altitude and speed, constant altitude with variable speed, variable altitude with constant speed, variable altitude with variable speed , random flight, and multi-UAV flight in same time. The main contribution of this work is to provide a flight dataset with state changes, which is absent in existing works. AMOVFLY-Dataset captures detailed flight dynamics and real-time environmental information, including wind speed, wind direction, position, altitude, speed, and energy consumption. AMOVFLY offers significant potential for advancing research in UAV-related time-series prediction and multi-UAV cooperation.

We validated AMOVFLY's application in the following work: 1) We enhanced the capability of Temporal Convolutional Network (TCN) in energy consumption prediction by demonstrating that training TCN on DJI and AMOV datasets improves its ability to support heterogeneous UAV platforms and adapt to diverse flight behaviors. 2) We found that that there is an urgent need for real-time updating mechanisms for UAV digital twin models, and that this dataset can maintain model performance and prevent degradation over time. 3) Additionally, AMOVFLY provides insights into UAV trajectory planning, revealing environmental impacts on flight paths through real-world flight discrepancies.(We have provided the details of these application experiments in our paper "AMOVFLY: Enabling Advanced UAV Modeling with The Comprehensive Flight Status Dataset", which has been submitted to TKDE and is currently under review.)

If you have any questions about how we collect the data or how to use it, please don't hesitate to get in touch with [email protected]. We hope AMOVFLY dataset will be useful for you!

Flight Data

Flight Data Statistics

We categorize and store various flight data in different folders according to different flight scenarios. Specifically, the FAFS folder contains data from drones flying at a fixed altitude and speed; the FAVS folder stores data from drones flying at a fixed altitude but with varying speeds; the VAFS folder holds data from drones flying at varying altitudes but with a fixed speed; the VAVS folder preserves data from drones flying at varying altitudes and speeds; and the Random folder keeps data from drones flying at random varying altitudes and speeds under manual control. Table below presents statistical data on the proportion of data from different scenarios, the aircraft used, flight durations, and other relevant information within the entire dataset.

Attention:The symbol "+" indicates that we will continue to supplement this type of data.

Flight Scenario	Flight Count	Related Aircraft	Flight Duration	Dataset Proportion
FAFS	33	UavY	318.5 mins	12.2%
FAVS	164	UavY, UavR, UavG	1741.4 mins	60.7%
VAFS	32+	UavY, UavR	293.5+ mins	11.8%
VAVS	27+	UavY, UavR	271.1+ mins	10.0%
Random	14+	UavY, UavR	140.1+ mins	5.1%
Multi-UAV	(25 pairs)+	(UavR, UavY) (UavY, UavG)	250.0 mins	/
Summary	270+	UavR, UavY, UavG	2764.6+ mins (46.1+ hours)	100%

Flight File Naming Rules

In this work, we keep the raw unprocessed acquisition data and the ready data that are directly used for the study after merging the wind speeds, which are saved in the subfolder raw_data and the subfolder with the same name as the parent folder, respectively. We summarise where these raw and ready data are saved, the naming convention, and the contents therein in in table below.

Folder	Sub-Folder	File	Content
Scenarios name	raw_data	[UAVcolor]_Wind_[PayloadParameter][AltitudeParameter][SpeedParameter]_[FlightNumber].csv	Raw wind speed and wind angle data
Scenarios name	raw_data	[UAVcolor]_[PayloadParameter][AltitudeParameter][SpeedParameter]_[FlightNumber]_date[MMDDHHMM]_b[BatteryCode]_[Collector].csv	Raw flight data
Scenarios name	Scenarios name	Uav[UAVcolor]_[PayloadParameter]AltitudeParameter][SpeedParameter]_[Flight Number].csv	Ready Data:flight data combined wind speed and wind angle

Ready Data Contents and Acquisition Methods

This subsection describes all the information contained in the readiness data and how this information is obtained, as shown in table below.

Variable	Unit	Description	Data Source
time	s	Time elapsed during flight	rostopic:/mavros/imu/data
wind_speed	m/s	Wind speed relative to the drone's flight	HY-SA256 Anemometer USB Serial Port
wind_angle	deg	Wind angle relative to the direction of drone's flight	HY-SA256 Anemometer USB Serial Port
air_pressure	Pa	Realtime air pressure during flight	rostopic:/mavros/imu/static_pressure
battery_voltage	V	System voltage measured immediately after the battery	rostopic:/mavros/battery
battery_current	A	System current measured immediately after the battery	rostopic:/mavros/battery
battery_remain	%	Remaining battery level	rostopic:/mavros/battery
gps_x;y	m	Position relative to the takeoff point	rostopic:/mavros/local_position/pose
gps_z	m	Altitude above the ground	rostopic:/mavros/local_position/pose
real_lat;_long	deg	Longitude/Latitude of the actual trajectory	rostopic:/mavros/global_position/raw/fix
aim_lat;_long	deg	Longitude/Latitude of the reference trajectory	republish rostopic:/current_waypoint
o_x; _y; _z; _w	quaternion	Aircraft orientation	rostopic:/mavros/imu/data
v_x; _y; _z	m/s	Ground speed	rostopic:/mavros/local_position/velocity_local
la_x; _y; _z	m/s²	Ground acceleration	rostopic:/mavros/local_position/velocity_local (check if correct topic for acceleration)
power	W	UAV battery output power (calculated as battery_voltage * battery_current).	Calculated from battery_voltage and battery_current values

Flight Information

In Flight_info.csv, we have provided supplementary explanations of the flight settings for each flight. This file records all external information related to the flight, including the scene setting for each flight (which is also the name of the folder containing the data files), takeoff time, flight route, data file name, UAV used, flight duration, battery number used, power consumption, total flight power, local wind speed, measured wind speed at the collection points, local atmospheric pressure, measured atmospheric pressure at the collection points, air density at the collection points, local temperature, local weather, and data collectors. The column name, descriptionand, and data source of each piece of information in the file are shown in the table below.

Column name	Description	Data Source
Data Dir	The folder name containing the flight data (file name corresponds to the flight scene)	Parsed from raw flight data name
Date time	The date and time when the flight started	Parsed from raw flight data name
Route	The reference flight route for this flight	Parsed from raw flight data name
FlightName	The file name of the flight data	Manually entered
Uav	The UAV used for this flight	Parsed from raw flight data name
Length(s)	The flight duration for this flight	Parsed from ready flight data file (the absolute difference between the first and last values in the "time" column for this flight)
BatteryName	The battery number used for this flight	Parsed from raw flight data name
BatteryCost	The total power consumption for this flight	Parsed from ready flight data file (the difference between the first and last power values for this flight)
AllPower	The total power used during this flight	Parsed from ready flight data file (the sum of all power data for this flight)
WindSpeed_station	Wind speed observation data from the nearest weather station at the corresponding time	Data requested from Visual Crossing Weather web API
WindSpeed_test	The average of the UAV's measured absolute wind speed data on the day	Average of the middle 50% of wind speed data in the corresponding date's wind speed file in the "wind" folder
AirPressure_station	Atmospheric pressure observation data from the nearest weather station at the corresponding time	Data requested from Visual Crossing Weather web API
AirPressure_test	The average atmospheric pressure during this flight	Parsed from ready flight data
Air Density	Air density	Calculated from AirPressure_station and WindSpeed_station data
Temperature	Temperature observation data from the nearest weather station at the corresponding time	Data requested from Visual Crossing Weather web API
Weather	Weather observation data from the nearest weather station at the corresponding time	Data requested from Visual Crossing Weather web API
Pick Man	The data collector for this flight	Manually entered

Multi-UAV Information

The Multi-UAV_infosheet.csv file records the corresponding data file names and takeoff times for each UAV during multi-UAV flights.The column name, and descriptionand in the file are shown in the table below.

Column Name	Description
Uav1	Name of the first UAV in multi-UAV flight
Uav1_time	Takeoff time of the first UAV
Uav2	Name of the second UAV in multi-UAV flight
Uav2_time	Takeoff time of the second UAV
Uav3	Name of the third UAV in multi-UAV flight
Uav3_time	Takeoff time of the third UAV

Flight Route

This dataset's flights involve a total of five different routes, and the aerial views and 3D simulations of these routes are stored in the "Route" folder.

Absolute Wind

In the "wind" folder, we have recorded the absolute wind condition information for the collection days, providing users with real-world environmental references. This is distinct from the relative wind condition information in the data files, which was collected during the UAV's flight along the reference trajectory at different speeds. Absolute wind conditions refer to wind data collected while the UAV is hovering at a fixed altitude. In this folder, we differentiate the absolute wind condition data from different collection days using the naming convention: [UAV Name]_wind_[Collection Time]_[Collection Altitude].csv. The "Collection Altitude" indicates the height at which the UAV hovered during data collection. For example, "102040" signifies that the UAV hovered at 10m, 20m, and 40m, while "20" indicates that the UAV only hovered at 20m to collect wind conditions.

Dataset Collectors and Contributors

The 270 flight data entries in the AMOVFLY dataset were collected by the following individuals. Their full names, email addresses, and associated organizations are listed in the table below:

Collector	Full name	Email	Organization
lmj or mj	Mengjie Lee	[email protected]	PhD student, Northwestern Polytechnical University
hfh	Fanghao Han	[email protected]	Master student, Northwestern Polytechnical University
ly	Yi Liu	[email protected]	Master student, Northwestern Polytechnical University
rcc	Chuncheng Ran	[email protected]	Master student, Northwestern Polytechnical University
ltc	Tianci Li	[email protected]	/