FUNMAP on Noetic

Does anyone know if FUNMAP in a good working state on Noetic?

I’m using the dev/noetic branch

I’m seeing a bunch of errors, so before I dig in and start filing issues, I wanted to ask here.

Hi @smth, FUNMAP should work fine with Noetic on the dev/noetic branch. Would you mind posting a log of the errors you see when running FUNMAP. I’d be happy to look into the issues you’re seeing.

There were a few other issues, that I think I got over just by upgrading (and then putting the realsense downgrade patch), including upgrading stretch_body and upgrading the latest apt updates.

The main issue now, seems to be that, after the first scan, the robot refuses to go anywhere further.
I tried two things:

  1. I give the “2D Nav Goal” command in the UI, with points nearby the robot with clear line of sight / no obstacles
  2. I press the \ key so that it goes and does further scans.

Neither seems to help, and instead in both cases, it shows an error in the console message: "Unable to find a path to the new head scan location."

Here’s what RViz looks like after the first scan.

Here are the logs: rosout.log · GitHub

Hi @smth, there are phantom points (yellow below) being included in the scan. It seems that they come from the robot including part of its own base in the scan. These points act as obstacles, preventing the robot from being able to find a valid path.


We’re able to replicate the issue internally, and I’m investigating fixes for the issue. Would you be able to perform a check of your current calibration by following these four commands in Stretch Calibration’s README, and report the total error you receive?

The fit error is high, but not high-enough to hit the 0.05 threshold:

total_error = 0.03806170965635635
[WARN] [1633101126.695582]: The fit error is high: 0.03806170965635635 > 0.03 (fit_error > fit_warning_threshold)

I re-calibrated the robot from scratch, and with the new calibration, the phantom points are gone, and the robot is navigating correctly


Awesome! Glad to hear it.