crowdAI / marLo

Multi Agent Reinforcement Learning using MalmÖ
MIT License
245 stars 46 forks source link

Boost.Python.ArgumentError #25

Open mamengyiyi opened 6 years ago

mamengyiyi commented 6 years ago

Hi, I installed the malmo few weeks ago and I could run all the .py in Python_Examples. But today I installed the marlo using "pip3 install -U marlo". And when I tested the example codes here [https://marlo.readthedocs.io/en/latest/usage/singleagent_example.html#example-code], it went wrong like these below: Boost.Python.ArgumentError: Python argument types in AgentHost.setObservationsPolicy(AgentHost, ObservationsPolicy) did not match C++ signature What should I do?

spMohanty commented 6 years ago

@mamengyiyi : Thats unusual. Can you also use the latest version of malmo ?

And if you are on linux, or osx, then you could try using the conda package directly as mentioned here : https://marlo.readthedocs.io/en/latest/installation.html

mamengyiyi commented 6 years ago

Hi, I use Anaconda now. But I have some questions. If I launch the agent in the "source activate marlo" environment, it will cause the error like below

[23:14:29] [Client thread/INFO] [STDOUT]: [net.minecraft.init.Bootstrap:printToSYSOUT:600]: #@!@# Game crashed! Crash report saved to: #@!@# /home/mayi/anaconda3/envs/marlo/Minecraft/run/./crash-reports/crash-2018-08-13_23.14.29-client.txt :runClient FAILED

FAILURE: Build failed with an exception.

  • What went wrong: Execution failed for task ':runClient'. Process 'command '/home/mayi/anaconda3/envs/marlo/bin/java'' finished with non-zero exit value 255

  • Try: Run with --stacktrace option to get the stack trace. Run with --info or --debug option to get more log output.

BUILD FAILED

And if I launch the client directly in the Minecraft folder, I can launch two clients. But when I run the multiagent example from the website, the two agents will appear in the same location and do nothing without looking into each others' eyes. Is that normal?