-
Notifications
You must be signed in to change notification settings - Fork 4
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Audit of topics to ensure no cheating allowed #172
Comments
Original comment by clalancette (Bitbucket: clalancette, GitHub: clalancette). As of today (May 19, 2017), here is the list of topics published/subscribed by the /gazebo node, along with the services in SRC (output from
|
Original comment by Steve Peters (Bitbucket: Steven Peters, GitHub: scpeters). I guess we could probably narrow it down further to topics published by or subscribed by the gazebo node. I think the IHMC controller doesn't have any secret knowledge. I think the main thing we are concerned about is gazebo. I forgot to mention services too; we may want to limit those. |
Original comment by Víctor López (Bitbucket: Victor Lopez). I agree with @scpeters that only the topics and services published or subscribed by gazebo should be blacklisted. And not all of them, obviously /clock and the srcsim topics are required. I don't believe that anything ihmc topic publishes anything on the simulation that would not be available on the real robot. |
Original comment by clalancette (Bitbucket: clalancette, GitHub: clalancette). I've now restricted my work to only look at topics that the /gazebo nodes pubs/subs/services. Looking at the publications, it all seems pretty reasonable, with the exception of:
Looking at the list of subscriptions, it looks fairly reasonable with the exception of:
Finally, looking at the list of services, the ones that are potentially problematic:
@chapulina , @scpeters , which of the above topics/services should we block? Once I have your input on these, I'll come up with a final list of topics/services that we need to block for now. |
Original comment by clalancette (Bitbucket: clalancette, GitHub: clalancette). All right, so with the feedback, the following should be blocked:
|
Original comment by Steve Peters (Bitbucket: Steven Peters, GitHub: scpeters). I wondered if we could just not use the |
Original comment by Steve Peters (Bitbucket: Steven Peters, GitHub: scpeters). I had forgotten that we also spawn valkyrie dynamically using the |
Original comment by Dirk Thomas (Bitbucket: Dirk Thomas, GitHub: dirk-thomas). If https://osrf-migration.github.io/srcsim-gh-pages/#!/osrf/srcsim/pull-requests/83/add-ros-subscriber-to-force-current/diff gets merged its topic needs to be disabled for the competition. |
Original comment by Víctor López (Bitbucket: Victor Lopez). Also all /controller_manager services need to be blacklisted. We asked at the beginning of the challenge (https://osrf-migration.github.io/srcsim-gh-pages/#!/osrf/srcsim/issues/51/custom-controller (#51)) if we could load our own controllers and the answer was no. So I believe that to be fair to all competitors who have adapted to use IHMC controller those services should be blacklisted. |
Original comment by Deanna Hood (Bitbucket: d_hood). @j-rivero added ros-simulation/gazebo_ros_pkgs#585 to disable most /gazebo topics and services which might be useful to you here. For ARIAC we needed a subset of them enabled (e.g.spawn_urdf_model, unpause physics), so we have a custom branch of gazebo_ros_packages being built in our dockerfiles. from a comment above it sounds like you might have the same requirements for SRC |
Original comment by clalancette (Bitbucket: clalancette, GitHub: clalancette). @d_hood Thanks. SRC has the same problem as ARIAC in that most of the topics/services want to be disabled, but not all of them. @j-rivero and I hammered out a proposal to have a blacklist of topics and services that can be specified at runtime, but neither of us has had time to implement it. |
Original comment by Jose Luis Rivero (Bitbucket: Jose Luis Rivero, GitHub: j-rivero). From today's gazebo meeting we agree on thinking that is too late to modify the runtime code for SRCSim (dry-run was last week). The approach to detect cheating is going to be different. I will mail you privately @clalancette |
Original report (archived issue) by clalancette (Bitbucket: clalancette, GitHub: clalancette).
This issue is to track an audit of the available topics in the SRC to ensure that none of the available topics can be used to cheat.
The text was updated successfully, but these errors were encountered: