Git Product home page Git Product logo

rl-texplore-ros-pkg's People

Contributors

asbroad avatar toddhester avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

rl-texplore-ros-pkg's Issues

Not Catkinized

The package is not catkinized which makes it unusable for ROS Hydro. 


Original issue reported on code.google.com by [email protected] on 6 Feb 2014 at 11:26

Re-Use a learned model

Hi,

I have a quick question concerning the re-use of a model. The scenario i am learning from takes a while to execute the actions from the rl_agent hence I would like to be able to somehow save the learned experience and re-use it later on.
From what i understand, this can be done using rl_msgs/RLEnvSeedExperience.msg message where before starting a new episode, I feed the past experience to the rl_agent which then takes it into consideration for planning in the next episode, is this correct?
Now concerning the SeedExperience message itself.

Message that contains a seed experience to initialize the model
float32[] from_state
int32     action
float32[] to_state
float32   reward
bool      terminal

I do not understand why the from_state and to_state are vectors? Is this because the rl_agent also accept experience from multi dimensional state spaces? In my case, the state space has 1 dimension. Would it mean that from_state and to_state would be one dimensional vectors for each seed?

I would appreciate if anyone could help.

Thanks,

Thibault

Parallel-UCT Memory Leak

Hi,

I am testing a simple example: the car velocity control with random set point. I have build all the code with no problems in ROS jade. I am running on a powerful MSI computer with 8 cores and 8GB of RAM (also 8GB of Swap Memory).

For the agent, I am running: rosrun rl_agent agent --agent texplore --lambda 0.6 --model m5tree --gamma 0.95 --planner parallel-uct --prints true --actrate 10

For the env, I am running: rosrun rl_env env --env carrandom --lag --stocstic --prints

As you can see, in this image:

parallel-uct

The memory occupation is continuously increasing. When it fills up the memory, the swap memory starts to be filled. After swap is full, Ubuntu directly kills the process and I can't keep on simulating.

Is this a known issue? Any idea? (Any "push_back" or "new" statement in the code which I should check?)

Thanks in advance.

Error when running make

Ran 'catkin_make' under top-level dir, then cd into src/ and tried running 'make' and got this error:

$ make

[  0%] Built target _rl_msgs_generate_messages_check_deps_RLStateReward
[  0%] Built target std_msgs_generate_messages_lisp
[  0%] Built target _rl_msgs_generate_messages_check_deps_RLAction
[  0%] Built target _rl_msgs_generate_messages_check_deps_RLEnvDescription
[  0%] Built target _rl_msgs_generate_messages_check_deps_RLEnvSeedExperience
[  0%] Built target _rl_msgs_generate_messages_check_deps_RLExperimentInfo
[  5%] Built target rl_msgs_generate_messages_lisp
[  5%] Built target std_msgs_generate_messages_py
[ 12%] Built target rl_msgs_generate_messages_py
[ 12%] Built target std_msgs_generate_messages_nodejs
[ 17%] Built target rl_msgs_generate_messages_nodejs
[ 17%] Built target std_msgs_generate_messages_eus
[ 24%] Built target rl_msgs_generate_messages_eus
[ 24%] Built target std_msgs_generate_messages_cpp
[ 29%] Built target rl_msgs_generate_messages_cpp
[ 29%] Built target rl_msgs_generate_messages
Scanning dependencies of target agentlib
[ 30%] Building CXX object rl_agent/CMakeFiles/agentlib.dir/src/Agent/DiscretizationAgent.cc.o
In file included from /home/dustin/code_projects/rl-texplore-ros-pkg/src/rl_agent/include/rl_agent/DiscretizationAgent.hh:4,
                 from /home/dustin/code_projects/rl-texplore-ros-pkg/src/rl_agent/src/Agent/DiscretizationAgent.cc:1:
/home/dustin/code_projects/rl-texplore-ros-pkg/src/rl_common/include/rl_common/Random.h:1129:23: error: ‘constexpr’ needed for in-class initialization of static data member ‘const float Random::_F’ of non-integral type [-fpermissive]
 1129 |    static const float _F    = 1. / _M;
      |                       ^~
make[2]: *** [rl_agent/CMakeFiles/agentlib.dir/build.make:63: rl_agent/CMakeFiles/agentlib.dir/src/Agent/DiscretizationAgent.cc.o] Error 1
make[1]: *** [CMakeFiles/Makefile2:1443: rl_agent/CMakeFiles/agentlib.dir/all] Error 2
make: *** [Makefile:141: all] Error 2

Solution:

change line 1129 of rl-texplore-ros-pkg/src/rl_common/Random.h to remove 'static' to be:

const float _F = 1. / _M;

After this change, running make worked just fine.

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.