File size: 3,159 Bytes
197258d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 |
extends VehicleBody3D
class_name Cart
var acceleration : float = 350
@onready var max_velocity = acceleration / mass * 40
@onready var ai_controller: AIController3D = $AIController3D
@export var destination: Node3D
@export var destination2: Node3D
@export var item: Item
var requested_acceleration: float
var initial_position: Vector3
var times_restarted: int
var item_collected: int
func get_normalized_velocity():
return linear_velocity.normalized() * (linear_velocity.length() / max_velocity)
func _ready():
initial_position = position
ai_controller.init(self)
func reset():
item_collected = 0
times_restarted += 1
position = Vector3(0, 0, randf_range(-8, 8))
rotation = Vector3.ZERO
reset_item()
linear_velocity = Vector3.ZERO
angular_velocity = Vector3.ZERO
pass
func reset_item():
var item_position : Vector3 = Vector3(0, 20, randf_range(-0.0, 0.0))
item.position = item_position
item.rotation = Vector3.ZERO
item.linear_velocity = Vector3.ZERO
item.angular_velocity = Vector3.ZERO
item.apply_central_force(Vector3(0.0, 0.0, randf_range(-70.0, 70.0)))
item.force_update_transform()
item.sleeping = false
item.set_category(randi_range(0, 1))
func _physics_process(delta):
reset_if_needed()
update_reward()
if (ai_controller.heuristic != "human"):
engine_force = (requested_acceleration) * acceleration
else:
engine_force = (int(Input.is_key_pressed(KEY_UP)) - int(Input.is_key_pressed(KEY_DOWN))) * acceleration
restart_if_outside_boundaries()
pass
func restart_if_outside_boundaries():
if (position.y < -2 or abs(position.z) > 10):
ai_controller.reward -= 1.0
ai_controller.needs_reset = true
ai_controller.done = true
func reset_if_needed():
if ai_controller.needs_reset:
reset()
ai_controller.reset()
## The reward function uses a simple form of curriculum learning, where initially
## a shaped reward tells the agent to move toward the item horizontally until it collects it,
## and after some episodes, only the sparse reward for collecting and delivering the item rewards.
func update_reward():
if times_restarted < 50:
if not item_collected:
ai_controller.reward -= 0.00001 * (item.position.z - position.z)
else:
if item.item_category == 0:
ai_controller.reward -= 0.00001 * (destination.position.z - position.z)
else:
ai_controller.reward -= 0.00001 * (destination2.position.z - position.z)
if item_collected:
if (position.distance_to(destination.position) < destination.scale.z):
if item.item_category == 0:
ai_controller.reward += 1.0
else:
ai_controller.reward -= 1.0
reset_item()
elif (position.distance_to(destination2.position) < destination2.scale.z):
if item.item_category == 1:
ai_controller.reward += 1.0
else:
ai_controller.reward -= 1.0
reset_item()
func _on_item_area_body_entered(body):
ai_controller.reward += 1.0
item_collected = 1
func _on_item_area_body_exited(body):
item_collected = 0
## If the item falls to the ground, gives a negative reward and resets the item
func _on_item_body_entered(body: PhysicsBody3D):
if body.get_collision_layer_value(3):
ai_controller.reward -= 0.05
reset_item()
|