updated on 24 Sept

algorithm/.ipynb_checkpoints/Untitled-checkpoint.ipynb

@@ -0,0 +1,6 @@
+{
+ "cells": [],
+ "metadata": {},
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

algorithm/FLP_Testing.py

@@ -0,0 +1,158 @@
+# -*- coding: utf-8 -*-
+"""
+Created on Sun Jun 19 12:25:32 2022
+
+@author: Shimraz
+"""
+
+import gym
+import gym.envs
+import pandas as pd
+from stable_baselines3 import DDPG
+import numpy as np
+import matplotlib.pyplot as plt
+import itertools
+import os
+import scipy.interpolate 
+import imageio
+from PIL import Image
+os.environ["KMP_DUPLICATE_LIB_OK"]="TRUE"
+
+class Test_FLP_Env:
+    def __init__(self,env_name,training_timesteps,ckpt_name_prefeix,
+                    log_path):
+        """
+        Description : 
+            Training class for the agent.
+            Here the agent is trained and saved with inbetween checkpoints.
+            Log is also saved in a csv file.
+
+        Input : 
+            env_name : str
+                Name of the Env to be tested on 
+
+            training_timesteps : int
+                Number of steps agent has trained
+
+            save_freq : int
+                After this many timesteps, model will be saved in between total training
+
+            ckpt_name_prefeix : str
+                Name prefix during checkpoint for the saved model
+
+            log_path : str
+                Folder where the log details are saved
+
+
+        """
+        training_timesteps = training_timesteps
+        ckpt_name_prefeix = ckpt_name_prefeix
+        log_path = log_path
+        env = gym.make(env_name,mode='rgb_array',instance='P6')
+
+        log_dir = log_path + "_ttsteps_"+ str(training_timesteps)                 
+
+
+        load_file = log_dir+"/FLP_Training_50000_steps.zip"              
+
+        # model = DDPG.load(load_file, env = env)
+        model = DDPG.load(load_file, env = env)
+        print(load_file)
+        self.Test(env, model, training_timesteps,load_file)
+
+
+
+    def Test(self,env, model,  train_steps, load_file):
+        """
+        Description : 
+            This function run the test file given the model and env
+
+        Input :
+            env : Env
+                Env to be tested on
+            model : DDPG Agent
+                DDPG agent used to predict the pressure
+        """
+        experiment_results={}
+        # global model
+        model = model
+        print("here")
+        # print("model", model)
+        for ts in range(train_steps):
+            ts = int(ts)
+            print(ts)
+            fig, (ax1,ax2) = plt.subplots(2,1)
+
+            obs = env.reset()
+            start_cost = getattr(env,"last_cost") #wrap_env.get_attr("last_cost")[0]
+            print(start_cost)
+            rewards = []
+            mhc = []
+            images = []
+            gain = 0
+            gains = []
+            actions = []
+            done = False
+            while done != True:
+                action, _states = model.predict(obs, deterministic = True)
+                actions.append(action)
+                obs, reward, done, info = env.step(action)
+
+                gain += reward
+                img =  env.render(mode='rgb_array')
+                rewards.append(reward)
+                # print(info)
+                mhc.append(info['mhc'])
+                gains.append(gain)
+                images.append(img)
+            print(mhc)
+            print(rewards)
+            final_cost = mhc[-1]
+            print("finalcost",final_cost)
+            cost_saved = final_cost-start_cost
+            cost_saved_rel = 1-(start_cost/final_cost)
+            print(cost_saved, cost_saved_rel, '%')
+            experiment_results[ts]=[cost_saved, cost_saved_rel]
+            ax1.plot(rewards)
+            ax2.plot(mhc)
+            print(actions)
+            imageio.mimsave(f'gifs/test_env.gif', [np.array(img.resize((200,200),Image.NEAREST)) for i, img in enumerate(images) if i%2 == 0], fps=29)
+            plt.show()
+            env.close()
+            del model
+        y = np.array([i for i in experiment_results.values()])
+        print(experiment_results, y)
+        print(train_steps, abs(y[:,0]))
+        fig1, axs = plt.subplots(1,1)
+        axs.plot(train_steps,abs(y[:,0]),)
+        plt.title("Plot flp")
+        print("end")
+
+
+
+if __name__ == '__main__':
+    """
+    Description :
+        Main file to test and plot the agent's prediction.
+
+    Parameters:
+
+        training_timesteps : int
+            Number of steps agent has trained
+
+        ckpt_name_prefeix : str
+            Name prefix during checkpoint for the saved model
+
+        log_path : str
+            Folder where the log details are saved
+
+        env_name : str
+            Name of the Env to be tested on
+    """   
+    training_timesteps = 10000   # Set accordingly
+    ckpt_name_prefeix = 'FLP_Training'
+    log_path = r'./FLP_Training/FLP_Training_v0'
+    #Name of the environment 
+    env_name = 'ofp-v1'
+
+    testing = Test_FLP_Env(env_name,training_timesteps,ckpt_name_prefeix,log_path)

algorithm/FLP_Training/FLP_Training_v0_ttsteps_50000/monitor.csv

@@ -0,0 +1,103 @@
+#{"t_start": 1661381832.9951506, "env_id": "ofp-v1"}
+r,l,t
+-1,1,0.135298
+-1,1,0.143295
+-1,1,0.1453
+-1,1,0.150295
+-1,1,0.154294
+-1,1,0.161295
+-1,1,0.166294
+-1,1,0.169295
+-1,1,0.176294
+-1,1,0.180295
+-1,1,0.183294
+-1,1,0.192294
+-1,1,0.196297
+-1,1,0.201295
+-1,1,0.207295
+-1,1,0.211294
+-1,1,0.218295
+-1,1,0.224299
+-1,1,0.229295
+-1,1,0.232294
+-1,1,0.235302
+-1,1,0.239295
+-1,1,0.241294
+-1,1,0.246294
+-1,1,0.252294
+-1,1,0.260294
+-1,1,0.274294
+-1,1,0.277294
+-1,1,0.289295
+-1,1,0.291295
+-1,1,0.294743
+-1,1,0.310371
+-1,1,0.310371
+-1,1,0.310371
+-1,1,0.310371
+-1,1,0.310371
+-1,1,0.325993
+-1,1,0.325993
+-1,1,0.325993
+-1,1,0.325993
+-1,1,0.325993
+-1,1,0.325993
+-1,1,0.341618
+-1,1,0.341618
+-1,1,0.357242
+-1,1,0.357242
+-1,1,0.357242
+-1,1,0.357242
+-1,1,0.372868
+-1,1,0.372868
+-1,1,0.372868
+-1,1,0.372868
+-1,1,0.388492
+-1,1,0.388492
+-1,1,0.404118
+-1,1,0.404118
+-1,1,0.404118
+-1,1,0.404118
+-1,1,0.404118
+-1,1,0.419743
+-1,1,0.419743
+-1,1,0.419743
+-1,1,0.419743
+-1,1,0.435366
+-1,1,0.435366
+-1,1,0.435366
+-1,1,0.435366
+-1,1,0.450992
+-1,1,0.450992
+-1,1,0.450992
+-1,1,0.466617
+-1,1,0.466617
+-1,1,0.466617
+-1,1,0.466617
+-1,1,0.482242
+-1,1,0.482242
+-1,1,0.497869
+-1,1,0.497869
+-1,1,0.497869
+-1,1,0.513493
+-1,1,0.513493
+-1,1,0.513493
+-1,1,0.529118
+-1,1,0.529118
+-1,1,0.529118
+-1,1,0.529118
+-1,1,0.544741
+-1,1,0.544741
+-1,1,0.544741
+-1,1,0.544741
+-1,1,0.544741
+-1,1,0.560367
+-1,1,0.560367
+-1,1,0.560367
+-1,1,0.575993
+-1,1,0.575993
+-1,1,0.575993
+-1,1,0.575993
+-1,1,0.591619
+-1,1,0.591619
+-1,1,0.60744