Greetings!

I have a rosbag with several topics and need help with one of them. My OS is Ubuntu 18 and Melodic install for ROS-1. I need to parse the values of x, y, and z from one topic that apears to be an object when I read the topic messages.

Normally this is straight forward. Rephrasing my question is "How do I read messages from the topic such that I can parse the values from a structure inside it?

Why do I know it is a structure? I can pull the topic from the bag and look at the cvs with pandas.

I can successfully do this in Matlab using their ros toolbox. I have clients who don't want to install ros or buy a MAtlab license but have a need to see the data from the ros bag. They could run a python script. That's where I could use your help.

The topic giving me issues is 'global_path'. I chose to use bagpy although I think importing rosbag would work as well.

The basic steps I followed were:

• read bag

  b = bagreader('/home/bags/flight12022-08-22-13-09-30.bag')

• check available topics

  b.topic_table

      Topics         Types                   Message Count  Frequency

• 0 /globalpath >> plannermap_interfaces/Plan >> 2 >>0.001536
• 1 /particlefiltersvisualizer >> visualization_msgs/MarkerArray >> 820 >>0.999696
• 2 /planner/planningrequest >> plannermap_interfaces/PlanningRequest >> 2 >> 0.001537
• 3 /searchbelief >> visualizationmsgs/Marker >> 1399 >> 2.000637
• 4 /targetprojectionnode/filteredtargets >> plannermap_interfaces/FilteredTargets >> 343 >>0.496940
• 5 /targetpropagation/particlefiltersbelief >> targetpropagation/ParticleFiltersBelief >> 820 >> 0.999689

• pull messages from topic /globalpath >>>data = b.messagebytopic('/globalpath')
• print("File saved: {}".format(data)) :: creates a csv of global_path topic

• result: /home//bags/flight12022-08-22-13-09-30/global_path.csv

• import pandas as pd

dfplan = pd.readcsv(data)

df_plan :: shows the table with 2 rows and 6 columns.'plan' is column 6

df_plan.shape :: shows (2,6) This is 2 rows and 6 columns

• an example of showing the first column of the csv, 'Time'.

• Note the type of the first is float64. So are columns 2 thru 5

df_plan['Time']

0 1.661199e+09

1 1.661200e+09

Name: Time, dtype: float64

• Next we see what type is the 6th column.. >>df_plan['plan'] >>> 0 [stamp: \n seq: 10\n stamp: \n secs: 0\n ...
  

1 [stamp: \n seq: 8\n stamp: \n secs: 0\n ...

Name: plan, dtype: object

• The data type of column 6 is an object with label 'plan''.

• the resultant cvs is a single column with many embedded fields

  • How can I format this 6th column? Is it when I first read the bag or when I extract the topic?
  • How do I read this column to extract the x, y, z positions inside the object?
  • in Matlab, I have a line to say the data type is a structure.
  • is there an equivalent using bagpy's bagreader or a method in rosbag?

• Note the MAtlab use of 'readMessages' to specify a data format of 'struct'

bag = rosbag('/path/to/rosbag file');

bSel = select(bag,'Topic','/global_path');

msgStructs = readMessages(bSel,'DataFormat','struct');

msgStructs{1} ::this references plan 1 of 2

msgStructs{1}.Plan_

• for n = 1:11
  • xEast(n) = msgStructs{1}.Plan_(n).Position(1).Position.X;
  • yNorth(n)= msgStructs{1}.Plan_(n).Position(1).Position.Y;
  • zUP(n) = msgStructs{1}.Plan_(n).Position(1).Position.Z;
• end

This has been challenging. I'm willing to share my bag or csv's. I tried rosbag, bagpy and pandas. I may have to parse the one-column result from the csv of the object represented by the 6th column. . I most humbly request any insight to extract the data of x, y, z from the dataframe object. Perhaps an inner loop and read each message as a row? I can provide files or do tests that help resolve this.. Kindest Regards