Sun 07 April 2013

Extracting raw data from ros topics

Posted by ankur in Research (664 words, approximately a 3 minute read)

Recently, I needed to extract raw data that we had collected from the PR2 robot. It's easy to collect data from topics, using the rosbag tool. You can collect data from multiple topics in one bag at a time.

Here's an example command that collects data from the robot:

rosbag record -b 1024 "/joint_states" -e "(/.*)_scan$" "torso_lift_imu/data" "/base_odometry/odom" "/camera/rgb/image_rect_color" "/camera/depth_registered/image_rect" --duration=1m --split -O /removable/recordings/20130403_1218_navtest_data.bag

The rosbag then looks like this:

[ankur@localhost  bag-test]$ rosbag info 20130403_1218_navtest_data_0.bag
path:        20130403_1218_navtest_data_0.bag
version:     2.0
duration:    58.8s
start:       Apr 03 2013 15:40:18.91 (1364964018.91)
end:         Apr 03 2013 15:41:17.68 (1364964077.68)
size:        3.3 GB
messages:    17035
compression: none [3295/3295 chunks]
types:       nav_msgs/Odometry      [cd5e73d190d741a2f92e81eda573aca7]
             sensor_msgs/Image      [060021388200f6f0f447d0fcd9c64743]
             sensor_msgs/JointState [3066dcd76a6cfaef579bd0f34173e9fd]
             sensor_msgs/LaserScan  [90c7ef2dc6895d81024acba2ac42f369]
topics:      /base_odometry/odom                   5516 msgs    : nav_msgs/Odometry
             /base_scan                            1176 msgs    : sensor_msgs/LaserScan
             /camera/depth_registered/image_rect   1640 msgs    : sensor_msgs/Image
             /camera/rgb/image_rect_color          1653 msgs    : sensor_msgs/Image
             /joint_states                         5875 msgs    : sensor_msgs/JointState
             /tilt_scan                            1175 msgs    : sensor_msgs/LaserScan
[ankur@localhost  bag-test]$

You can also filter out individual topics into individual bags with the tool:

[ankur@localhost  bag-test]$ rosbag filter --help
Usage: rosbag filter [options] INBAG OUTBAG EXPRESSION

EXPRESSION can be any Python-legal expression.

The following variables are available:
 * topic: name of topic
 * m: message
 * t: time of message (t.secs, t.nsecs)

Filter the contents of the bag.

  -h, --help            show this help message and exit
                        Python expression to print for verbose debugging. Uses
                        same variables as filter-expression
[ankur@localhost  bag-test]$

Various other tools are available to visualize the data. With groovy, you have rqt_bag

Since I needed the raw messages, I had to hack up a python script using the rosbag API.


# Copyright 2010 Ankur Sinha
# Author: Ankur Sinha
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# GNU General Public License for more details.
# You should have received a copy of the GNU General Public License
# along with this program.  If not, see <>.
# File :

import rosbag
import sys
import os
import pickle

# Global variable for input file name

def run():
    Main run method. Calls other helper methods to get work done

    if len(sys.argv) != 2:
        sys.stderr.write('[ERROR] This script only takes input bag file as argument.n')
        inputFileName = sys.argv[1]
        print "[OK] Found bag: %s" % inputFileName

        bag = rosbag.Bag(inputFileName)
        topicList = readBagTopicList(bag)

        while True:
            if len(topicList) == 0:
                print "No topics in list. Exiting"
            selection  = menu(topicList)

            if selection == -92:
                print "[OK] Printing them all"
                for topic in topicList:
                    extract_data(bag, topic, inputFileName)
            elif selection == -45:
                topic = topicList[selection]
                extract_data(bag, topic, inputFileName)


def extract_data (bag, topic, inputFileName):
    Spew messages to a file

        topic -> topic to extract and print to txt file

    outputFileName = os.path.splitext(os.path.split(inputFileName)[1])[0] + topic.replace("/","-") + ".txt"
    print "[OK] Printing %s" % topic
    print "[OK] Output file will be called %s." % outputFileName

    outputFh = open(outputFileName, "w")

    for topic, msg, t in bag.read_messages(topics=topic):

    print "[OK] DONE"

def menu (topicList):
    Print the user menu and take input

        topicList: tuple containing list of topics

        selection: user selection as integer

    i = 0
    for topic in topicList:
        print '[{0}] {1}'.format(i, topic)
        i = i+1
    if len(topicList) > 1:
        print '[{0}] Extract all'.format(len(topicList))
        print '[{0}] Exit'.format(len(topicList) + 1)
        print '[{0}] Exit'.format(len(topicList))

    while True:
        print 'Enter a topic number to extract raw data from:'
        selection = raw_input('>>>')
        if int(selection) == len(topicList):
            return -92 # print all
        elif int(selection) == (len(topicList) +1):
            return -45 # exit
        elif (int(selection) < len(topicList)) and (int(selection) >= 0):
            return int(selection)
            print "[ERROR] Invalid input"

def readBagTopicList(bag):
    Read and save the initial topic list from bag
    print "[OK] Reading topics in this bag. Can take a while.."
    topicList = []
    for topic, msg, t in bag.read_messages():
        if topicList.count(topic) == 0:
            topicList.append (topic)

    print '{0} topics found:'.format(len(topicList))
    return topicList

if __name__ == "__main__":

I've hosted it on my github repository. It'll be easier to read/download it from there if you need to.