As a part of a project in WAVELab on system identification and modeling of a quadrotor for Multi-Step prediction, a dataset was gathered. This page contains information about this dataset. The results of a Black-box and a Grey-box modeling are reported in [1, 2].
(Maintained by Nima Mohajerin)
Format MATLAB .mat Size 238.1 MB
Download here: AscTec_Pelican_Flight_Dataset
The AscTec Pelican is a light weight quadrotor. It is equipped with a real-time autopilot board coupled with an onboard computer running Ubuntu OS and communicates with the autopilot board via a UART connection. The ROS Indigo software running a suitable ROS node is used to collect the motor speeds and Inertial Measurement Unit (IMU) measurements. The vehicle is operated in an indoor environment by an expert pilot using a Futaba T7C remote control.
Download this documentation.
The dataset contains 54 flights covering a wide range of regimes. The flights are stored as MATLAB cells with the following fields:
- len: The number of samples in the flight
- Pos: The position measurement (x, y, z) from Vicon, a 3-by-len matrix
- Euler: The orientation measurement (roll, pitch, yaw) from Vicon, a 3-by-len matrix
- Motors: The motors actual speed as reported by the ROS node, 4-by-len matrix
- Motors_CMD: The commanded motor speeds as reported by the ROS node, a 4-by-len matrix
- Vel: The calculated velocity (v_x, v_y, v_z) by applying a numerical difference on the position measurement, a 3-by-(len-1) matrix
- pqr: The calculated body rates (p, q, r) by applying a numerical difference on the Euler measurements and transferring them into the body frame, a 3-by-(len-1) matrix
If you use this dataset, kindly cite our papers:
[1] Deep Learning a Quadrotor Dynamic Model for Multi-Step Prediction, N Mohajerin, M Mozifian, SL Waslander IEEE International Conference on Robotics and Automation (ICRA), 2018.
[2] Multistep Prediction of Dynamic Systems With Recurrent Neural Networks, N Mohajerin, SL Waslander IEEE transactions on neural networks and learning systems, 2019.