Rosbag Inspection Survey

General
, ,
1

Hello all,

We at AWS RoboMaker are interested in feedback from the community on logs and metrics in robotics development to help us learn more about the needs, pain-points and usage seen in this space. We’d greatly appreciate it if you can fill out the survey below.

Click here for the survey

It takes less than 5 minutes to complete, and we will keep the link open for a couple weeks at-least. We will be providing the feedback received back in this thread shortly after closing the survey.

12 Likes
2

Please help us collect data via this survey, it only takes a few minutes.

Thanks in advance!

1 Like
3

I am just leaving this here… :wink:

image
image804×302 13.8 KB

I am spamming almost as much as The Construct :stuck_out_tongue_winking_eye:

9 Likes
4

How much longer will the survey be open?

5

@mkhansen We are planning to keep the survey open till August 7th.

1 Like
6

hahaha @facontidavide I filled out the survey and did that exact same thing. I wrote in plotjuggler as Other. Just now seeing your response, too. I imagine you won’t be the only one.

7

Here is how we manage (clean up older than 2 weeks) rosbags:

find /var/log/bagfiles -type d -mtime +14 | xargs rm -rf

Otherwise we fill the SSD hard drive on our production field units. We pull when we’re back in the office at HQ over ethernet to a storage array, and would love to utilize some AWS goodness (like S3) one day.

1 Like
8

Thanks for putting together the survey! I filled in the form. 2 cents about probably minor points about the survey itself:

  • Many questions ask about “robot”. My team does not use ROS to control what one calls as “robot” (one with some sort of actuators), and I assume there are others in the similar situation, so I was almost inclined to respond to those questions as “we don’t use ROS to control robots”, but I assumed the intent by “robot” was something like “the system you develop on top of ROS”.
  • I felt many questions assume the surveyee uses rosbag as a main logging/replaying tool, which I bet makes a bit difficult to answer for some people. Maybe this is out of scope of this survey and the project though, it might be beneficial to add more options and sections for those who chose not to use rosbag to describe why.
9

Care to elaborate a bit more? As any product manager will tell you survey’s are great, but narrative descriptions of user pain points are often just as useful.

10

Hey Charles,

The two weeks timeline is similar to what my experience was in building robots too, a few years ago. Thanks for the feedback.

We open-sourced ROS packages that makes it easy to upload files and specifically rosbags to S3, you might want to check that out!


1 Like
11

@130s , You are correct on your first point. You could respond to the survey with a system that you use ROS/rostopics and rosbags with, and not necessarily a robot with actuators in mind.

Echoing @Katherine_Scott, we are interested in hearing you use case and pain points. Do let us know!

12

Thank you for those, we have a story in our tracker specifically for researching and doing a development spike on using these libraries. Keep up the great work!

2 Likes
13

7th - 4th = 3 days left to participate in the survey! (I’m good at math!)
:robot:

1 Like
14

I think I’ll need to see more evidence before I can believe such a bold claim.

2 Likes
15

Last call! Survey ends today!

Thanks again for your help!

:robot:

16

it would be nice to get the results of the survey, since the community contributed to it :wink:

6 Likes
17

Yes, we will share results once we have had a few days to review and analyze them

18

Hello all,

Thank you all for the responses for this survey. We had 88 responses! You can find the results in this doc rosbag_survey_results.pdf (225.4 KB)

For a quick overview, some of the findings were as follows

  • 85% (75/88) of surveyed customers use rosbags

  • 91% (80/88) of surveyed customers inspect rosbags from simulation

  • 60% (54/88) of surveyed customers inspect rosbags from simulation

    • 51% (45/88) for single test troubleshooting and
    • 24% (21/88) for CI/CD troubleshooting, with near complete overlap between the groups.

  • Majority of customers (91%) still use rviz. Common reasons for why they currently use rviz and rqt:

    • Ease of use : “It gets the job done with minimal effort“, “speed and convenience” (26%)
    • Inertia : “comes with ROS”, “Most engineers already know those tools and using alternatives slows down development and increases development costs” (35%)

  • Rosbags from robots:

    • Rosbag storage time – majority of customers delete rosbags within a week or never delete until they run out of diskspace
    • Typical bag size – pretty even and depends on use case, but large chunk between 0.1 – 2 gb/min
    • Bandwidth – trends towards high bandwidth, but varying stability
1 Like