GridEnvironment3.py

#next step

def step(self, action):
            
            action = self.actions
            reward_dict = {'up':10, 'down':10, 'left':10, 'right':10, 'begin':0, 'obstacle':-5}
        
            next_state = (self.agent_state[0]+self.actions_pos_dict[actions][0],
                          self.agent_state[1]+self.actions_pos_dict[actions][1] )
            
            if action == "up":
                return (self.observation, reward_dict.value[0], action)
            
            elif action == "down":
                return (self.observation, reward_dict.value[1], action)
            elif action == "right":
                return (self.observation, reward_dict.value[2], action)
            
            elif action == "left":
                return (self.observation, reward_dict.value[3], action)
            else:
                return (self.observation, reward_dict.value[4])#the agent begins from start position
            
            
            #out of bounds condition
            if next_state[0] < 0 or next_state[0] >= self.grid_shape[0]:
                return (self.observation, "This state is out of bounds")
            
            elif next_state[1] <0 or next_state[1] >= self.grid_shape[1]:
                
                return (self.observation, "This state is out of bounds")
                
            #obstacle avoidance
            iterations =10
            for i in range(iterations):
                for j in range(obstacles):
                    if next_state[i] == self.obstacle_mask[j]:
                        start == [0,0]
                        next_state == start
                        
                        return (self.observation, next_state, reward_dict.value[5])
                    #condition for starting again
                    elif next_state == "begin":
                        self.observation = self.reset()
                        return (self.observation, reward_dict.value[4])
                    
                       
        def reset(self):
            self.agent_state = copy.deepcopy(self.start_state)
            self.current_map = copy.deepcopy(self.initial_map)
            self.observation = self.grid_map_observation(self.initial_map)

            self.render()
            return self.observation