Datasets:
Formats:
imagefolder
Size:
< 1K
File size: 5,050 Bytes
c6c8030 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 |
extends CharacterBody2D
# Declare member variables here. Examples:
# var a = 2
# var b = "text"
const pad = 100
const WIDTH = 1280
const HEIGHT = 720
const MAX_FRUIT = 10
var _bounds := Rect2(pad,pad,WIDTH-2*pad,HEIGHT-2*pad)
@export var speed := 500
@export var friction = 0.18
var _velocity := Vector2.ZERO
var _action = Vector2.ZERO
var _heuristic = "player"
@onready var fruit = $"../Fruit"
@onready var raycast_sensor = $"RaycastSensor2D"
@onready var walls := $"../Walls"
@onready var colision_shape := $"CollisionShape2D"
var fruit_just_entered = false
var just_hit_wall = false
var done = false
var best_fruit_distance = 10000.0
var fruit_count = 0
var n_steps = 0
var MAX_STEPS = 20000
var needs_reset = false
var reward = 0.0
func _ready():
raycast_sensor.activate()
reset()
func _physics_process(delta):
n_steps +=1
if n_steps >= MAX_STEPS:
done = true
needs_reset = true
if needs_reset:
needs_reset = false
reset()
return
var direction = get_direction()
if direction.length() > 1.0:
direction = direction.normalized()
# Using the follow steering behavior.
var target_velocity = direction * speed
_velocity += (target_velocity - _velocity) * friction
set_velocity(_velocity)
move_and_slide()
_velocity = velocity
update_reward()
if Input.is_action_just_pressed("r_key"):
reset()
func reset():
needs_reset = false
fruit_just_entered = false
just_hit_wall = false
#done = false
fruit_count = 0
_velocity = Vector2.ZERO
_action = Vector2.ZERO
position = _calculate_new_position()
spawn_fruit()
# position.x = randf_range(_bounds.position.x, _bounds.end.x)
# position.y = randf_range(_bounds.position.y, _bounds.end.y)
# fruit.position.x = randf_range(_bounds.position.x, _bounds.end.x)
# fruit.position.y = randf_range(_bounds.position.y, _bounds.end.y)
best_fruit_distance = position.distance_to(fruit.position)
n_steps = 0
func _calculate_new_position(position: Vector2=Vector2.ZERO) -> Vector2:
var new_position := Vector2.ZERO
new_position.x = randf_range(_bounds.position.x, _bounds.end.x)
new_position.y = randf_range(_bounds.position.y, _bounds.end.y)
if (position - new_position).length() < 4.0*colision_shape.shape.get_radius():
return _calculate_new_position(position)
var radius = colision_shape.shape.get_radius()
var rect = Rect2(new_position-Vector2(radius, radius),
Vector2(radius*2, radius*2)
)
for wall in walls.get_children():
#wall = wall as Area2D
var cr = wall.get_node("ColorRect")
var rect2 = Rect2(cr.get_position()+wall.position, cr.get_size())
if rect.intersects(rect2):
return _calculate_new_position()
return new_position
func get_direction():
if done:
_velocity = Vector2.ZERO
return Vector2.ZERO
if _heuristic == "model":
return _action
var direction := Vector2(
Input.get_action_strength("move_right") - Input.get_action_strength("move_left"),
Input.get_action_strength("move_down") - Input.get_action_strength("move_up")
)
return direction
func set_action(action):
_action.x = action["move"][0]
_action.y = action["move"][1]
func reset_if_done():
if done:
reset()
func get_obs():
var relative = fruit.position - position
var distance = relative.length() / 1500.0
relative = relative.normalized()
var result := []
result.append(((position.x / WIDTH)-0.5) * 2)
result.append(((position.y / HEIGHT)-0.5) * 2)
result.append(relative.x)
result.append(relative.y)
result.append(distance)
var raycast_obs = raycast_sensor.get_observation()
result.append_array(raycast_obs)
return {
"obs": result,
}
func update_reward():
reward -= 0.01 # step penalty
reward += shaping_reward()
func zero_reward():
reward = 0.0
func get_reward():
return reward
func shaping_reward():
var s_reward = 0.0
var fruit_distance = position.distance_to(fruit.position)
if fruit_distance < best_fruit_distance:
s_reward += best_fruit_distance - fruit_distance
best_fruit_distance = fruit_distance
s_reward /= 100.0
return s_reward
func set_heuristic(heuristic):
self._heuristic = heuristic
func get_obs_space():
# typs of obs space: box, discrete, repeated
return {
"obs": {
"size": [len(get_obs()["obs"])],
"space": "box"
}
}
func get_action_space():
return {
"move" : {
"size": 2,
"action_type": "continuous"
}
}
func get_done():
return done
func set_done_false():
done = false
func spawn_fruit():
fruit.position = _calculate_new_position(position)
best_fruit_distance = position.distance_to(fruit.position)
func fruit_collected():
fruit_just_entered = true
reward += 10.0
fruit_count += 1
spawn_fruit()
# if fruit_count > MAX_FRUIT:
# done = true
func wall_hit():
done = true
reward -= 10.0
just_hit_wall = true
reset()
func _on_Fruit_body_entered(body):
fruit_collected()
func _on_LeftWall_body_entered(body):
wall_hit()
func _on_RightWall_body_entered(body):
wall_hit()
func _on_TopWall_body_entered(body):
wall_hit()
func _on_BottomWall_body_entered(body):
wall_hit()
|