Loading
Cookies help us deliver our services. By using our services, you agree to our use of cookies. Learn more

crowdAI is shutting down - please read our blog post for more information

NIPS 2017: Learning to Run

Reinforcement learning environments with musculoskeletal models


Completed
2154
Submissions
631
Participants
88660
Views

What difficulty on the leaderboard?

Posted by NNAISENSE almost 2 years ago

What is the difficulty we are faced when submitting solutions? Is it 1 or 2? Should be 2 probably but the gifs on the leaderboard suggest it is 1 (no large obstacles).

I am asking because I cannot see it on my PC when submitting. The visualizer says: simbody-visualizer: received Shutdown message. Goodbye. Xlib: sequence lost (0x100a5 > 0xab) in reply type 0x12! [xcb] Unknown request in queue while dequeuing [xcb] Most likely this is a multi-threaded client and XInitThreads has not been called [xcb] Aborting, sorry about that. simbody-visualizer: ../../src/xcb_io.c:179: dequeue_pending_request: Assertion `!xcb_xlib_unknown_req_in_deq' failed. [2017-08-16 16:03:07,005] POST http://grader.crowdai.org:1729/v1/envs/ {"env_id": "Run", "token": "05e9d84f485cfeca02016dfd4d18163c", "version": "1.4.1"}

Posted by spMohanty  almost 2 years ago |  Quote

Hi @NN,

The obstacles are actually the small blue balls that you see in the visualisations. And the grader runs on difficulty=2.

The error that you mention, seems to be because of a misconfigured opensim environment. Also are you sure that the server you are trying to run this on has a display attached ?

Cheers, Mohanty

Posted by NNAISENSE  almost 2 years ago |  Quote

Hi @NN,

The obstacles are actually the small blue balls that you see in the visualisations.

Right but the when I run the simulation locally on difficulty 2 some of the obstacles are quite large and on the gifs all of them are equal and small but maybe this is a coincidence.

And the grader runs on difficulty=2.

The error that you mention, seems to be because of a misconfigured opensim environment. Also are you sure that the server you are trying to run this on has a display attached ?

Sure. I can run the visualization when running the local version (i.e., RunEnv).

Posted by spMohanty  almost 2 years ago |  Quote

Hi @NN,

Yeah the size and location of the obstacles depends on the SEED used to instantiate the environment. The good and generalisable models should be able to happily hop/walk/run through all the obstacles. :smile:

Mohanty