DeepTerrainRL

Intro

Source code for the paper: Terrain-Adaptive Locomotion Skills using Deep Reinforcement Learning

Setup

This section covers some of the steps to setup and compile the code. The software depends on many libraries that need to be carefully prepared and placed for the building and linking to work properly.

Linux

1. Caffe (http://caffe.berkeleyvision.org/installation.html) Specific version (https://github.com/niuzhiheng/caffe.git @ 7b3e6f2341fe7374243ee0126f5cad1fa1e44e14) sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler sudo apt-get install --no-install-recommends libboost-all-dev sudo apt-get install libgflags-dev libgoogle-glog-dev liblmdb-dev sudo apt-get install libatlas-base-dev

   In the instruction to make and build Caffe uncomment the CPU only line

   # CPU-only switch (uncomment to build without GPU support).
   CPU_ONLY := 1

   Or if on Windows https://github.com/initialneil/caffe-vs2013

2. Boost

3. OpenCV

4. BulletPhysics

5. CUDA Package Manager Installation Install repository meta-data When using a proxy server with aptitude, ensure that wget is set up to use the same proxy settings before installing the cuda-repo package. $ sudo dpkg -i cuda-repo-__.deb Update the Apt repository cache $ sudo apt-get update Install CUDA $ sudo apt-get install cuda

6. Json_cpp (https://github.com/open-source-parsers/jsoncpp)

7. Eigen (http://eigen.tuxfamily.org/index.php?title=Main_Page)

8. bits sudo apt-get install gcc-4.9-multilib g++-4.9-multilib

Windows

Runing The System

After the system has been build there are two executable files that server different purposes. The **TerrainRL** program is for visually simulating the a controller and **TerrainRL_Optimize** is for optimizing the parameters of some controller.

Examples: To simulate a controller/character ./Terr