Add new Edit Delete History Source

Data Analysis

During the operation in the ELISSA laboratory or the simulation, the topic data can be recorded using ROS bag. This data can be converted to csv files which usually are plotted with python. This process is described by an example on this page.

Table of Content:

Data Analysis
- Data Conversion
- Data Visualization

Data Conversion

tbd

Data Visualization

Dependencies

The following packages are needed and imported first:

import pandas as pd
from datetime import datetime, timedelta
from scipy.spatial.transform import Rotation
import matplotlib.pyplot as plt
import os

Loading the Data

Work in Progress

To visualize the data, a python class is created. As an input, it receives the foldername containing the csv files. It is checked, that the folder exists. This ist happening in the __init__ method:

class Data:

    def __init__(self, foldername, delta_min=0.15, delta_max=2, convert_time_man=True, show_plots=False, save_plots=True):

        print(f'Reading data from folder {foldername}')

        # Get the data of the force sensors
        self.force_files = []
        self.force_files.append(f'{foldername}/__force_sens_00__dock_force.csv')
        self.force_files.append(f'{foldername}/__force_sens_01__dock_force.csv')
        self.force_files.append(f'{foldername}/__force_sens_02__dock_force.csv')
        self.force_files.append(f'{foldername}/__force_sens_03__dock_force.csv')

        # Get the data for the ff states
        self.state_files = []
        self.state_files.append(f'{foldername}/chaser__state.csv')
        self.state_files.append(f'{foldername}/chaser__state_goal.csv')
        self.state_files.append(f'{foldername}/target__state.csv')

The data is read from the csv file using the pandas package. The data is saved in easy to handle dataframes. These dataframes are saved in a list. During the loading, the separator must be specified.

 # Read the data from the csv files
        self.force_data = []
        self.force_data_raw = []
        for file in self.force_files:
            self.force_data.append(pd.read_csv(file, sep=','))
            self.force_data_raw.append(pd.read_csv(file, sep=','))
        self.state_data = []
        for file in self.state_files:
            self.state_data.append(pd.read_csv(file, sep=','))

Processing the Data

Next, the loaded data is processed. Multiple steps are necessary, to put the available data in a efficiently usable format. First, the timestamps are converted to a datetime format for easier handling:

# Convert the time into the datetime format
        for i, element in enumerate(self.force_data):
            tmp_list = []
            for time in self.force_data[i]['Timestamp']:
                tmp_list.append(datetime.fromtimestamp(time))
            self.force_data[i]['Timestamp'] = tmp_list
        for i, element in enumerate(self.state_data):
            tmp_list = []
            for time in self.state_data[i]['Timestamp']:
                tmp_list.append(datetime.fromtimestamp(time))
            self.state_data[i]['Timestamp'] = tmp_list

To get better structured data, the columns must be renamed. As at the end all data should be merged in one dataframe, we must ensure that no columns have the same name.

        # Rename and sort the columns
        if plot_x or plot_vel:
            self.data[0] = self.clean_state(self.data[0], app='Goal_')
            self.data[1] = self.clean_state(self.data[1])
        if plot_wrench:
            self.data[2] = self.clean_wrench(self.data[2], app='Exc_')
        if plot_ctl_wrench:
            self.data[3] = self.clean_wrench(self.data[3], app='Ctl_')
        self.data[4] = self.clean_veh_state(self.data[4])