Drone Motion Primitive Dataset

Drone technology is one of the largest tackled fields in today’s world, as it can range from pure enjoyment of drone racing to medical use and fighting crime. Several teams are interested in developing improved human machine interfaces for operating drones. This dataset is a collection of different motion primitives commanded using a PS3 joystick to control an Ardrone on Gazebo. This has been conducted using ROS Melodic on Ubuntu 18.04. The PS3 six-axis joystick was used to collect this data, and it was connected to the Gazebo drone by using the ArDrone Helpers libraries.This dataset contains 4 different motion primitives which are: Line, Helix, Sine, and Arc. The testing was conducted on each primitive alone and using a stop indicator, which is specified for each primitive, at the end of each primitive. The stop indicators are as follows:· X for Sine· Circle for Helix· Square for Line· Triangle for ArcThen a collection of every two primitives consecutively was also conducted (i.e. line to helix, arc to line, helix to sine …) using the same stop indictors at the end of every corresponding primitive. Each test has around 300 to 500 samples. It is also important to note that our data is sample and hold, meaning that if the joystick’s analog stick is in a certain position for a few seconds, then it will not send any new samples for the velocities.The data was extracted into a .YAML file, as continuous commands from the joystick which were used to move the drone in Gazebo. They are then parsed (using python 2.7) to extract the velocities of the drone along the x y z axes as well as the pressed buttons (x or circle or square or triangle) into a .CSV file.This parsed data is further processed (using python3) into another .CSV file containing all the velocities as arrays, a histogram to get the average number of samples for every sequence, and a .RTF file which contains some analysis for each primitive or collection of primitives (number of samples taken, min and max sequence size, mean, median, and deviation).N.B: for the sine primitive, the data has been extracted into two categories: one where the sine was along the x-y axes and another where it was along the x-z or y-z axes. The github link for the two codes used is: https://github.com/justinelhalabi/PS3-Joystick-Command-Parsing-into-CSV

无人机技术是当前全球重点研究与应用的核心领域之一，其应用范畴从纯粹的无人机竞速娱乐，延伸至医疗应用与犯罪防控等诸多方向。目前已有多个研发团队致力于开发更优化的人机交互界面（Human Machine Interface, HMI）以实现无人机操控。本数据集收录了基于PS3摇杆在Gazebo仿真平台中操控Ardrone无人机的各类运动基元（motion primitives）指令数据，本次数据采集基于Ubuntu 18.04系统与ROS Melodic版本完成。本次数据采集使用PS3六轴摇杆完成，通过ArDrone Helpers库实现摇杆与Gazebo仿真平台内无人机的连接。

本数据集包含四类不同的运动基元，分别为直线（Line）、螺旋线（Helix）、正弦曲线（Sine）与圆弧（Arc）。实验分别对单类运动基元进行测试，每类基元末端均设有对应专属的停止标识，各类基元对应的停止标识如下：
· 正弦曲线（Sine）对应按键X
· 螺旋线（Helix）对应按键Circle
· 直线（Line）对应按键Square
· 圆弧（Arc）对应按键Triangle

随后，本数据集还收录了连续组合两类运动基元的测试数据（例如直线转螺旋线、圆弧转直线、螺旋线转正弦曲线等），每类组合的对应基元末端仍采用前述专属停止标识。单组测试的样本量约为300至500条。

需特别说明的是，本数据集采用采样保持（sample and hold）机制：若摇杆的模拟摇杆保持某一位置数秒，则不会再发送新的速度采样数据。首先将Gazebo平台中操控无人机的摇杆连续指令提取为.YAML格式文件；随后使用Python 2.7对该文件进行解析，将无人机沿x、y、z轴的速度数据与触发的按键（X、Circle、Square或Triangle）导出为.CSV格式文件。

之后，使用Python 3对解析后的.CSV文件进行二次处理，生成三类输出文件：一是将所有速度数据整理为数组形式的.CSV文件；二是用于统计每类序列平均样本量的直方图文件；三是.RTF格式的分析报告文件，其中包含每类基元或基元组合的相关统计指标，包括采集样本数、序列长度的最值、均值、中位数与标准差。

注：针对正弦曲线基元，本数据集将其数据分为两类：一类为沿x-y轴运动的正弦曲线数据，另一类为沿x-z或y-z轴运动的正弦曲线数据。本次实验所用的两段代码的GitHub链接为：https://github.com/justinelhalabi/PS3-Joystick-Command-Parsing-into-CSV