player construction
The aim of the player class is to construct players with different ways of interacting with the environment and generate observations with controlled randomness. This is done using giving probabilities to a player.
basic player
A basic player has these references
- id
- observation
- random generator
- type of player
- reward array
- punishment array
other than these parameters a seed was added where it will always a certain way of randomness every time calls the class in order to keep the robustness.
observation spectrum
this was constructed into 3 main parts - gaze - distance - task completion
gaze
this represent do the player maintain the gaze with the robot.
| value | observation |
|---|---|
| 0 | not maintaining the observation correctly or out of focus |
| 1 | maintaining the observation with the robot |
distance
as save as the above in the gaze
| value | observation |
|---|---|
| 0 | not maintaining the observation correctly or out of focus |
| 1 | maintaining the observation with the robot |
task
task was divided into 4 main parts
| value | observation |
|---|---|
| 0 | best performance |
| 1 | average performance |
| 2 | week performance |
| 3 | no performance (not responding) |
creating a player
in order to create a player player class should be called.
Player(self,player_id:int=0, player_type =0 , prob = [[0.1, 0.9],
[0.1, 0.9],
[0.8, 0.1, 0.1, 0.0]])
from players.player3 import Player as Player
player1 = Player()
ret = player1.generate(prob = [[0.1, 0.9],
[0.1, 0.9],
[0.8, 0.1, 0.1, 0.0]] )
print(ret)
setting up epsilon values for each player can be done, in order to setup the epsilon values
player = Player..epsilon_fn(frame_idx:int = None)