What tools have people developed and use for rosbag analysis?
I know of rqt_bag (which is an absolutely awesome tool), which allows you to browse the topics in a concise timeline, and using that in conjunction with rqt_plot to plot data. However, it's still a little cumbersome to extract data for more detailed analysis.
I've created a small package using numpy that will parse a rosbag given a set of topic paths (consisting of fields and optionally indices, much like extracting data for plotting) and return simple multi-variable data sets for ease of plotting and analysis, which includes interpolating data sets so that they are on a time scale (with the same spacing). This allows basic errors to be computed. If you also published the same data, you can use this to potentially see delays between the different publishers. You can always run the conversion, then save the final datasets to Matlab to plot and examine as well. Really easy to do in Spyder.
Will upload the package in a bit.
EDIT: Googled and saw Pandas from this presentation: https://speakerdeck.com/nzjrs/managing-complex-experiments-automation-and-analysis-using-robot-operating-system Looks interesting, will start looking at this.
Originally posted by eacousineau on ROS Answers with karma: 174 on 2013-10-16
Post score: 3