MXNET-Scala Playing Flappy Bird Using Deep Reinforcement Learning

MXNet-scala module implementation of DQN to Play Flappy Bird.

Based on: https://github.com/li-haoran/DRL-FlappyBird

result:

Setup

Tested on Ubuntu 14.04

Requirements

• sbt 0.13 http://www.scala-sbt.org/
• Mxnet https://github.com/dmlc/mxnet

steps

1, compile Mxnet with CUDA, then compile the scala-pkg，doc： https://github.com/dmlc/mxnet/tree/master/scala-package

2, under the Mxnet-Scala/DRLFlappyBird folder

 mkdir lib;

3, copy the compiled mxnet-full_2.11-INTERNAL.jar into lib folder;

4, run sbt then compile the project

Training

using the script scripts/run.sh:

#### training #####

java -Xmx4G -cp $CLASS_PATH \
	FlappyBirdDQN \
	--gpu $GPU \
	--save-model-path $SAVE_MODRL_PATH \
	--resources-path $RESOURCES_PATH

Running with pretrain models

using the script scripts/run.sh, comment the training part and uncomment the folllowing line:

#### resume training ####

RESUME_MODRL_PATH=$ROOT/models/pretrain-model/network-dqn_mx46000.params

java -Xmx4G -cp $CLASS_PATH \
   FlappyBirdDQN \
  	--gpu $GPU \
  	--save-model-path $SAVE_MODRL_PATH \
  	--resources-path $RESOURCES_PATH \
  	--resume-model-path $RESUME_MODRL_PATH

Have fun !!