Project_Navigation_RL_Agent_Inference.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\palin\\Anaconda3\\lib\\site-packages\\statsmodels\\tools\\_testing.py:19: FutureWarning: pandas.util.testing is deprecated. Use the functions in the public API at pandas.testing instead.\n",
      "  import pandas.util.testing as tm\n"
     ]
    }
   ],
   "source": [
    "from unityagents import UnityEnvironment\n",
    "import numpy as np\n",
    "import torch\n",
    "import os\n",
    "import seaborn as sns\n",
    "import time\n",
    "from collections import namedtuple\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from DeepQAgent.Agent import Agent as DQNAgent\n",
    "from DeepQAgent.Agent_Priority import Agent as DQN_Prioritized_Agent\n",
    "from DeepQAgent.Agent_Priority import ExpBuffer\n",
    "\n",
    "from DeepQAgent.QNetwork import QNetwork"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>Load & instantiate the RL Environment</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:unityagents:\n",
      "'Academy' started successfully!\n",
      "Unity Academy name: Academy\n",
      "        Number of Brains: 1\n",
      "        Number of External Brains : 1\n",
      "        Lesson number : 0\n",
      "        Reset Parameters :\n",
      "\t\t\n",
      "Unity brain name: BananaBrain\n",
      "        Number of Visual Observations (per agent): 0\n",
      "        Vector Observation space type: continuous\n",
      "        Vector Observation space size (per agent): 37\n",
      "        Number of stacked Vector Observation: 1\n",
      "        Vector Action space type: discrete\n",
      "        Vector Action space size (per agent): 4\n",
      "        Vector Action descriptions: , , , \n"
     ]
    }
   ],
   "source": [
    "# Load the Environment\n",
    "env = UnityEnvironment(file_name=\"./Navigation_Environment/Banana.exe\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "brain_name = env.brain_names[0]\n",
    "brain = env.brains[brain_name]\n",
    "\n",
    "action_size = brain.vector_action_space_size       # get the size of the action space\n",
    "env_info = env.reset(train_mode=True)[brain_name] # reset the environment\n",
    "state = env_info.vector_observations[0]            # get the current state\n",
    "state_size = len(state)\n",
    "score = 0                                          # initialize the score\n",
    "seed = 0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>Define H-params & instantiate RL Agent object</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "IS_PRIORITIZED_REPLAY = False #True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# total number of Episodes for Training Agent\n",
    "num_episodes = 100\n",
    "\n",
    "# Define the Agent\n",
    "\n",
    "agent = None\n",
    "\n",
    "if IS_PRIORITIZED_REPLAY:\n",
    "    # Agent instantiation with Prioritized Experience Replay\n",
    "    agent = DQN_Prioritized_Agent( state_size, action_size, seed,\n",
    "                                   torch.device('cuda:0' if torch.cuda.is_available() else 'cpu:0')\n",
    "                                 )\n",
    "else:\n",
    "    # Agent instantiation without Prioritized Experience Replay\n",
    "    agent = DQNAgent( state_size, action_size, seed )\n",
    "\n",
    "epsilon_thresh = 0.1 # 0.1\n",
    "epsilon_decay = 0.99 # 0.999995 # find a decay factor so that epsilon decays each timestep to close to 0 in One episode\n",
    "episodic_epsilon_decay = 0.985\n",
    "epi_epsilon = 1/(episodic_epsilon_decay) # The first Episode should start from 1.0 -> we'll be decaying over Episodes too\n",
    "episode_score = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0., 0., 0., ..., 0., 0., 0.],\n",
       "       [0., 0., 0., ..., 0., 0., 0.],\n",
       "       [0., 0., 0., ..., 0., 0., 0.],\n",
       "       [0., 0., 0., ..., 0., 0., 0.]])"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# define container to store the Action-heatmap\n",
    "# There are 300 timesteps in each episode\n",
    "# ~ define a Heatmap container which has 'Action-Space' number of rows & 'episode-timesteps' equivalent\n",
    "# number of columns\n",
    "\n",
    "action_hmap = np.zeros((action_size, 300))\n",
    "\n",
    "action_hmap"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h4>Load stored Agent DQN model</h4>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<All keys matched successfully>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load from check point if any\n",
    "agent.qnetwork_local.load_state_dict( torch.load('./DeepQAgent/Vanilla_Q_Agent_QLocal_Score_Thresh_13.4.pth')['state_dict'] )\n",
    "agent.qnetwork_target.load_state_dict( torch.load('./DeepQAgent/Vanilla_Q_Agent_QTarget_Score_Thresh_13.4.pth')['state_dict'] )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>Run Inference on the Environment for the Loaded Agent Model</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Episode: 100 ~ Tstep: 301| Score: 8.0                                                               \n",
      "Total Elapsed Simulation time: 149.365 seconds | 2.49 minutes\n"
     ]
    }
   ],
   "source": [
    "start_time = time.time()\n",
    "\n",
    "try: \n",
    "    for i in range(num_episodes):\n",
    "        \"\"\"\n",
    "            At the beginning of the Episode ->\n",
    "                1> Initialise the episode_score = 0\n",
    "                2> Reset the Environment State\n",
    "        \"\"\"\n",
    "        score = 0                                         # Initialise Agent's episode score\n",
    "        inspect_score = 0                                 # Initilaise Avg Score over 'inpect_episodes'\n",
    "        env_info = env.reset(train_mode=True)[brain_name] # reset the environment\n",
    "        state = env_info.vector_observations[0]           # get the curent state\n",
    "        \n",
    "        # epsilon = 1.0 at first Episode -> decay episodic epsilon over episodes\n",
    "        epi_epsilon = epi_epsilon*episodic_epsilon_decay  \n",
    "        epi_epsilon = 0.0 if epi_epsilon < epsilon_thresh else epi_epsilon # threshold to stop Agent action Exploration\n",
    "        \n",
    "        epsilon = epi_epsilon # init epsilon for one Episode which wil decay over timesteps in an Episode\n",
    "        \n",
    "        done = False\n",
    "        episode_start_time = time.time()\n",
    "        timestep = 1\n",
    "        yellow, blue = 0, 0\n",
    "\n",
    "        # Keep running the environment simulation until the Terminal State has been reached\n",
    "        # EPISODE RUN\n",
    "        while not done:\n",
    "            \n",
    "            action, q_estimate = None, None\n",
    "            \n",
    "            if IS_PRIORITIZED_REPLAY:\n",
    "                # Agent WITH Prioritized Action Replay\n",
    "                action, q_estimate = agent.act(state, epsilon) # Agent Action\n",
    "            else:\n",
    "                # Agent without any Prioritized Action Replay\n",
    "                action = agent.act(state, epsilon) # action = agent.act(state)\n",
    "            \n",
    "            action = action.astype('int32')\n",
    "            \n",
    "            # Update the Action Heatmap for the specific timestep in the Episode\n",
    "            action_hmap[action][timestep-1] += 1\n",
    "            \n",
    "            env_info = env.step(action)[brain_name] # send Agent action to the Environment\n",
    "\n",
    "            # Obtain:> 'next_state', 'reward', 'done' params for the Agent-Environment interaction\n",
    "            next_state = env_info.vector_observations[0]   \n",
    "            reward = env_info.rewards[0]                   \n",
    "            done = env_info.local_done[0]     \n",
    "            \n",
    "            # Update the timestep-score\n",
    "            score += reward\n",
    "\n",
    "            # State-transition: update current state to next state\n",
    "            state = next_state\n",
    "\n",
    "            # Epsion (Greedy parameter) decay over timesteps & capping to low_thresh of 0.1\n",
    "            epsilon = epsilon * epsilon_decay\n",
    "            epsilon = 0.0001 if epsilon < epsilon_thresh else epsilon  \n",
    "\n",
    "            timestep += 1\n",
    "\n",
    "        # One EPISODE RUN has completed\n",
    "        episode_score.append(score)\n",
    "\n",
    "        log=\"Episode: {} ~ Tstep: {}| Score: {}\".format( i+1, timestep, score )\n",
    "        \n",
    "        log = log + \" \"*(100 - len(log))\n",
    "        print( log, end = \"\\r\" );# print(log)\n",
    "        \n",
    "except KeyboardInterrupt:\n",
    "    print(\"\\nKeyboardInterrupt exception -> Aborting RL Agent Train\")\n",
    "    \n",
    "end_time = time.time()\n",
    "simul_time = np.round(end_time-start_time, 3)\n",
    "\n",
    "print(\"\\nTotal Elapsed Simulation time: {} seconds | {} minutes\".format( simul_time, np.round(simul_time/60.0, 2) )\n",
    "     )\n",
    "\n",
    "# Close the Environment\n",
    "env.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h4>Get Heatmap for Agent Reward observed during Inference</h4>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x72 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "plt.figure(figsize=(20, 1))\n",
    "sns.heatmap(np.reshape(episode_score, (1, len(episode_score))), cmap=\"rocket_r\")\n",
    "\n",
    "plt.xlabel(\"Episodes\")\n",
    "plt.title(\"Agent Reward Points per Episode during Inference Run\")\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h4>Get Heatmap for Agent Action-Space Decisions per timestep in Episode observed during Inference</h4>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "plt.figure(figsize=(20, 6))\n",
    "sns.heatmap(action_hmap/num_episodes, cmap=\"icefire\")\n",
    "\n",
    "plt.xlabel(\"Timesteps in an Episode\")\n",
    "plt.ylabel(\"Agent Actions\")\n",
    "plt.title(\"Agent Action-Heatmap per Timestep in Episode\")\n",
    "\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}