Reinforcement Learning
stable-baselines3
English
Korean
ppo
continuous-control
mujoco
humanoid-v5
robotics
robot
bipedal-robot
neuromotion
Instructions to use hwihwalab/neuromotion-humanoid-v5-ppo with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- stable-baselines3
How to use hwihwalab/neuromotion-humanoid-v5-ppo with stable-baselines3:
from huggingface_sb3 import load_from_hub checkpoint = load_from_hub( repo_id="hwihwalab/neuromotion-humanoid-v5-ppo", filename="{MODEL FILENAME}.zip", ) - Notebooks
- Google Colab
- Kaggle
File size: 28,488 Bytes
87f07b2 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 | import os
import warnings
os.environ["PYGAME_HIDE_SUPPORT_PROMPT"] = "1"
warnings.filterwarnings("ignore")
import collections
import numpy as np
import pygame
from simulation_engine import SimulationEngine
from train_rl import RLTrainer
# ==========================================================
# Design Token System (Slate & Semantic Neon Palette)
# ==========================================================
COLOR_BG = (10, 14, 23) # Slate 950 (Deep Slate Navy)
COLOR_PANEL = (17, 24, 39) # Slate 900
COLOR_PANEL_HOVER = (30, 41, 59) # Slate 800
COLOR_PANEL_BORDER = (30, 41, 59) # Slate 800 (1px Precision Border)
COLOR_PANEL_SUB = (13, 18, 30) # Deep Sub-panel Surface
COLOR_DROPDOWN_BG = (15, 23, 42) # Dropdown Menu Background
# Semantic Accent Tokens
COLOR_CYAN = (6, 182, 212) # Cyan 500
COLOR_EMERALD = (16, 185, 129) # Emerald 500
COLOR_AMBER = (245, 158, 11) # Amber 500
COLOR_ROSE = (244, 63, 94) # Rose 500
COLOR_VIOLET = (139, 92, 246) # Violet 500
COLOR_MAGENTA = (217, 70, 239) # Fuchsia 500
# Typography Tokens (WCAG 4.5:1 High Contrast)
COLOR_TEXT_PRIMARY = (248, 250, 252) # Slate 50
COLOR_TEXT_SECONDARY = (148, 163, 184) # Slate 400
COLOR_TEXT_MUTED = (100, 116, 139) # Slate 500
class RobotAILabGUI:
"""NEUROMOTION 3.0 Master Telemetry Dashboard Main Class."""
def __init__(self, width=1280, height=820):
pygame.init()
pygame.font.init()
self.width = width
self.height = height
self.screen = pygame.display.set_mode((self.width, self.height))
pygame.display.set_caption("NEUROMOTION 3.0 // Humanoid-v5 Continuous Control Telemetry")
self.clock = pygame.time.Clock()
self.is_running = True
self.is_paused = False
self.render_3d = True
# Speed multiplier controls
self.speed_multiplier = 1
self.speed_options = [
(1, "1X Normal (1x)"),
(2, "2X Fast (2x)"),
(4, "4X Hyper (4x)"),
(8, "8X Ultra (8x)"),
]
# Disturbance force magnitude presets
self.push_options = [
(15.0, "Light Push (15 N)"),
(30.0, "Medium Push (30 N)"),
(50.0, "Heavy Force (50 N)"),
]
# Active dropdown state ('speed', 'push', None)
self.active_dropdown = None
# Visual feedback alert timers
self.push_alert_timer = 0
self.push_alert_text = "[ DISTURBANCE INJECTED ]"
self.save_alert_timer = 0
# Modern typography system font initialization
font_names = "segoeui,malgungothic,arial,helvetica"
self.font_brand = pygame.font.SysFont(font_names, 18, bold=True)
self.font_sub = pygame.font.SysFont(font_names, 12)
self.font_card_val = pygame.font.SysFont(font_names, 24, bold=True)
self.font_card_lbl = pygame.font.SysFont(font_names, 11, bold=True)
self.font_mono = pygame.font.SysFont("consolas,monospace,courier", 12)
self.font_small = pygame.font.SysFont(font_names, 11)
self.font_alert = pygame.font.SysFont(font_names, 14, bold=True)
# Initialize physics engine and PPO trainer
print("[TELEMETRY] Initializing MuJoCo Simulation & PPO Engine...")
self.sim = SimulationEngine(render_width=590, render_height=420)
self.trainer = RLTrainer(model_path="humanoid_ppo_model.zip")
# Real-time telemetry buffers
self.episode_count = 1
self.best_reward = 0.0
self.recent_rewards = collections.deque(maxlen=100)
self.recent_20_rewards = collections.deque(maxlen=20)
self.console_logs = collections.deque(maxlen=5)
self.console_logs.append("[SYSTEM] NEUROMOTION 3.0 Telemetry Online.")
# Interactive mouse button geometry definitions
self._init_buttons()
def _init_buttons(self):
"""Initialize top header interactive button bounds."""
self.btn_pause = pygame.Rect(630, 12, 130, 32)
self.btn_speed = pygame.Rect(770, 12, 125, 32)
self.btn_push = pygame.Rect(905, 12, 135, 32)
self.btn_save = pygame.Rect(1050, 12, 100, 32)
self.btn_load = pygame.Rect(1160, 12, 100, 32)
self.buttons = [
("pause", self.btn_pause),
("speed", self.btn_speed),
("push", self.btn_push),
("save", self.btn_save),
("load", self.btn_load),
]
def run(self):
"""Main execution telemetry loop."""
train_step_timer = 0
while self.is_running:
self._handle_events()
if not self.is_paused:
for _ in range(self.speed_multiplier):
obs = self.sim.observation
action = self.trainer.predict_action(obs, deterministic=False)
next_obs, reward, done, info = self.sim.step(action)
if done:
ep_rew = self.sim.episode_reward
ep_steps = self.sim.alive_steps
self.recent_rewards.append(ep_rew)
self.recent_20_rewards.append(ep_rew)
if ep_rew > self.best_reward:
self.best_reward = ep_rew
log_msg = f"[LOG] EP {self.episode_count:02d} // REW: {ep_rew:+06.1f} // {ep_steps:02d} STEPS"
self.console_logs.append(log_msg)
self.episode_count += 1
self.sim.reset()
train_step_timer += 1
if train_step_timer % 128 == 0:
self.trainer.step_learning(total_steps=128)
if self.push_alert_timer > 0:
self.push_alert_timer -= 1
if self.save_alert_timer > 0:
self.save_alert_timer -= 1
self._draw_gui()
pygame.display.flip()
self.clock.tick(60)
self.sim.close()
pygame.quit()
def _handle_events(self):
"""Handle keyboard/mouse events and dropdown menu interactions."""
mouse_pos = pygame.mouse.get_pos()
is_hovering_btn = any(rect.collidepoint(mouse_pos) for _, rect in self.buttons)
is_hovering_dd = self._is_hovering_dropdown(mouse_pos)
pygame.mouse.set_cursor(pygame.SYSTEM_CURSOR_HAND if (is_hovering_btn or is_hovering_dd) else pygame.SYSTEM_CURSOR_ARROW)
for event in pygame.event.get():
if event.type == pygame.QUIT:
self.is_running = False
elif event.type == pygame.MOUSEBUTTONDOWN and event.button == 1:
if self.active_dropdown:
handled = self._handle_dropdown_click(mouse_pos)
if handled:
continue
if self.btn_pause.collidepoint(mouse_pos):
self.is_paused = not self.is_paused
self.trainer.is_training_active = not self.is_paused
self.active_dropdown = None
elif self.btn_speed.collidepoint(mouse_pos):
self.active_dropdown = None if self.active_dropdown == "speed" else "speed"
elif self.btn_push.collidepoint(mouse_pos):
self.active_dropdown = None if self.active_dropdown == "push" else "push"
elif self.btn_save.collidepoint(mouse_pos):
self.active_dropdown = None
self._trigger_save()
elif self.btn_load.collidepoint(mouse_pos):
self.active_dropdown = None
self._trigger_load()
else:
self.active_dropdown = None
elif event.type == pygame.KEYDOWN:
self.active_dropdown = None
if event.key == pygame.K_SPACE:
self.is_paused = not self.is_paused
self.trainer.is_training_active = not self.is_paused
elif event.key == pygame.K_t:
self.render_3d = not self.render_3d
elif event.key == pygame.K_f:
cur_idx = [opt[0] for opt in self.speed_options].index(self.speed_multiplier) if self.speed_multiplier in [1, 2, 4, 8] else 0
next_idx = (cur_idx + 1) % len(self.speed_options)
self.speed_multiplier = self.speed_options[next_idx][0]
elif event.key == pygame.K_p:
self._trigger_push(30.0, "Medium (30 N)")
elif event.key == pygame.K_s:
self._trigger_save()
elif event.key == pygame.K_l:
self._trigger_load()
elif event.key == pygame.K_e or event.key == pygame.K_r:
self.sim.reset()
self.console_logs.append("[RESET] Episode environment reset.")
def _is_hovering_dropdown(self, mouse_pos) -> bool:
if self.active_dropdown == "speed":
menu_rect = pygame.Rect(self.btn_speed.x, self.btn_speed.bottom + 4, 155, len(self.speed_options) * 30 + 8)
return menu_rect.collidepoint(mouse_pos)
elif self.active_dropdown == "push":
menu_rect = pygame.Rect(self.btn_push.x, self.btn_push.bottom + 4, 165, len(self.push_options) * 30 + 8)
return menu_rect.collidepoint(mouse_pos)
return False
def _handle_dropdown_click(self, mouse_pos) -> bool:
if self.active_dropdown == "speed":
for i, (val, label) in enumerate(self.speed_options):
item_rect = pygame.Rect(self.btn_speed.x, self.btn_speed.bottom + 6 + i * 30, 155, 28)
if item_rect.collidepoint(mouse_pos):
self.speed_multiplier = val
self.active_dropdown = None
self.console_logs.append(f"[SPEED] Rate set to {val}X.")
return True
elif self.active_dropdown == "push":
for i, (force, label) in enumerate(self.push_options):
item_rect = pygame.Rect(self.btn_push.x, self.btn_push.bottom + 6 + i * 30, 165, 28)
if item_rect.collidepoint(mouse_pos):
self._trigger_push(force, label)
self.active_dropdown = None
return True
return False
def _trigger_push(self, force: float = 30.0, label: str = "30 N"):
self.sim.apply_push_disturbance(force)
self.push_alert_timer = 35
self.push_alert_text = f"[ DISTURBANCE: +{force:.1f} N ]"
self.console_logs.append(f"[TEST] Force injected (+{force:.0f} N)")
def _trigger_save(self):
self.trainer.save_model("humanoid_ppo_model.zip")
self.save_alert_timer = 40
self.console_logs.append("[SAVE] Checkpoint weights saved")
def _trigger_load(self):
if self.trainer.load_model("humanoid_ppo_model.zip"):
self.console_logs.append("[LOAD] Checkpoint weights loaded")
else:
self.console_logs.append("[ERROR] Model file not found")
def _draw_gui(self):
"""Render master dashboard and sub-panels."""
self.screen.fill(COLOR_BG)
self._draw_header()
self._draw_main_3d_panel(x=20, y=60, w=590, h=420)
self._draw_stat_cards(x=630, y=60, w=630, h=110)
self._draw_learning_diagnostics(x=630, y=185, w=630, h=120)
self._draw_actuator_torques(x=630, y=320, w=630, h=440)
self._draw_reward_curve(x=20, y=495, w=290, h=265)
self._draw_console_logs(x=320, y=495, w=290, h=265)
self._draw_footer()
# Topmost dropdown menu overlay with ambient drop shadow
self._draw_dropdown_menus()
def _draw_header(self):
"""Render top header navigation and controls."""
mouse_pos = pygame.mouse.get_pos()
mouse_down = pygame.mouse.get_pressed()[0]
# Brand identity title block
pygame.draw.rect(self.screen, COLOR_CYAN, (20, 15, 4, 26), border_radius=2)
title_surf = self.font_brand.render("NEUROMOTION // HUMANOID-V5", True, COLOR_TEXT_PRIMARY)
self.screen.blit(title_surf, (32, 14))
sub_surf = self.font_mono.render("PPO CONTINUOUS CONTROL", True, COLOR_TEXT_MUTED)
self.screen.blit(sub_surf, (325, 20))
# 1. Pause/Resume button
status_col = COLOR_AMBER if self.is_paused else COLOR_EMERALD
status_txt = "PAUSED" if self.is_paused else "PPO TRAINING"
self._draw_engineering_button(self.btn_pause, status_txt, status_col, mouse_pos, mouse_down, dot=True)
# 2. Speed multiplier dropdown button
speed_colors = {1: COLOR_CYAN, 2: COLOR_VIOLET, 4: COLOR_MAGENTA, 8: COLOR_AMBER}
cur_col = speed_colors.get(self.speed_multiplier, COLOR_CYAN)
speed_btn_txt = f"SPEED {self.speed_multiplier}X v"
self._draw_engineering_button(self.btn_speed, speed_btn_txt, cur_col, mouse_pos, mouse_down, is_active=(self.active_dropdown == "speed"))
# 3. Disturbance injection dropdown button
self._draw_engineering_button(self.btn_push, "PUSH FORCE v", COLOR_ROSE, mouse_pos, mouse_down, is_active=(self.active_dropdown == "push"))
# 4. Model checkpoint save button
self._draw_engineering_button(self.btn_save, "SAVE MODEL", COLOR_CYAN, mouse_pos, mouse_down)
# 5. Model checkpoint load button
self._draw_engineering_button(self.btn_load, "LOAD MODEL", COLOR_TEXT_SECONDARY, mouse_pos, mouse_down)
def _draw_engineering_button(self, rect: pygame.Rect, text: str, accent_color, mouse_pos, mouse_down: bool, dot: bool = False, is_active: bool = False):
"""Interactive button renderer with hover and active states."""
is_hover = rect.collidepoint(mouse_pos)
is_pressed = (is_hover and mouse_down) or is_active
draw_rect = rect.move(0, 1) if is_pressed else rect
bg_col = COLOR_PANEL_HOVER if (is_hover or is_active) else COLOR_PANEL
pygame.draw.rect(self.screen, bg_col, draw_rect, border_radius=6)
border_col = accent_color if (is_hover or is_active) else COLOR_PANEL_BORDER
pygame.draw.rect(self.screen, border_col, draw_rect, width=1, border_radius=6)
if dot:
pygame.draw.circle(self.screen, accent_color, (draw_rect.x + 15, draw_rect.centery), 4)
txt_surf = self.font_sub.render(text, True, COLOR_TEXT_PRIMARY if is_hover else accent_color)
self.screen.blit(txt_surf, (draw_rect.x + 26, draw_rect.centery - 7))
else:
txt_surf = self.font_sub.render(text, True, accent_color if is_hover else COLOR_TEXT_PRIMARY)
txt_rect = txt_surf.get_rect(center=draw_rect.center)
self.screen.blit(txt_surf, txt_rect)
def _draw_dropdown_menus(self):
"""Topmost overlay dropdown menu with ambient drop shadow."""
mouse_pos = pygame.mouse.get_pos()
if self.active_dropdown == "speed":
menu_w = 155
menu_h = len(self.speed_options) * 30 + 8
menu_rect = pygame.Rect(self.btn_speed.x, self.btn_speed.bottom + 4, menu_w, menu_h)
# Ambient drop shadow (4px translucent border)
shadow_s = pygame.Surface((menu_w + 8, menu_h + 8), pygame.SRCALPHA)
shadow_s.fill((0, 0, 0, 140))
self.screen.blit(shadow_s, (menu_rect.x - 4, menu_rect.y - 2))
pygame.draw.rect(self.screen, COLOR_DROPDOWN_BG, menu_rect, border_radius=6)
pygame.draw.rect(self.screen, COLOR_PANEL_BORDER, menu_rect, width=1, border_radius=6)
for i, (val, label) in enumerate(self.speed_options):
item_rect = pygame.Rect(self.btn_speed.x + 4, self.btn_speed.bottom + 6 + i * 30, menu_w - 8, 26)
is_hover = item_rect.collidepoint(mouse_pos)
is_selected = (self.speed_multiplier == val)
if is_hover or is_selected:
pygame.draw.rect(self.screen, COLOR_PANEL_HOVER, item_rect, border_radius=4)
col = COLOR_CYAN if is_selected else (COLOR_TEXT_PRIMARY if is_hover else COLOR_TEXT_SECONDARY)
txt_surf = self.font_sub.render(label, True, col)
self.screen.blit(txt_surf, (item_rect.x + 8, item_rect.y + 5))
elif self.active_dropdown == "push":
menu_w = 165
menu_h = len(self.push_options) * 30 + 8
menu_rect = pygame.Rect(self.btn_push.x, self.btn_push.bottom + 4, menu_w, menu_h)
# Ambient shadow layer
shadow_s = pygame.Surface((menu_w + 8, menu_h + 8), pygame.SRCALPHA)
shadow_s.fill((0, 0, 0, 140))
self.screen.blit(shadow_s, (menu_rect.x - 4, menu_rect.y - 2))
pygame.draw.rect(self.screen, COLOR_DROPDOWN_BG, menu_rect, border_radius=6)
pygame.draw.rect(self.screen, COLOR_ROSE, menu_rect, width=1, border_radius=6)
for i, (force, label) in enumerate(self.push_options):
item_rect = pygame.Rect(self.btn_push.x + 4, self.btn_push.bottom + 6 + i * 30, menu_w - 8, 26)
is_hover = item_rect.collidepoint(mouse_pos)
if is_hover:
pygame.draw.rect(self.screen, COLOR_PANEL_HOVER, item_rect, border_radius=4)
col = COLOR_ROSE if is_hover else COLOR_TEXT_PRIMARY
txt_surf = self.font_sub.render(label, True, col)
self.screen.blit(txt_surf, (item_rect.x + 8, item_rect.y + 5))
def _draw_main_3d_panel(self, x, y, w, h):
"""Main 3D simulation viewport with telemetry overlay."""
panel_rect = pygame.Rect(x, y, w, h)
pygame.draw.rect(self.screen, COLOR_PANEL, panel_rect, border_radius=8)
if self.render_3d:
sim_surf = self.sim.render_surface()
self.screen.blit(sim_surf, (x, y))
else:
off_msg = self.font_sub.render("3D RENDERING SUSPENDED (PRESS T TO ENABLE)", True, COLOR_TEXT_MUTED)
self.screen.blit(off_msg, (x + w // 2 - 140, y + h // 2))
pygame.draw.rect(self.screen, COLOR_PANEL_BORDER, panel_rect, width=1, border_radius=8)
# Slim info bar with episode and step counters
self._draw_glass_pill(x + 12, y + 12, 105, 30, "EPISODE", str(self.episode_count), COLOR_CYAN)
self._draw_glass_pill(x + 125, y + 12, 105, 30, "ALIVE STEPS", str(self.sim.alive_steps), COLOR_EMERALD)
self._draw_glass_pill(x + 238, y + 12, 135, 30, "NOISE SCALE", f"sigma = {self.trainer.exploration_noise:.2f}", COLOR_AMBER)
# Disturbance injection HUD alert popup
if self.push_alert_timer > 0:
alert_s = pygame.Surface((w - 30, 36), pygame.SRCALPHA)
alert_s.fill((244, 63, 94, 210))
pygame.draw.rect(alert_s, (255, 255, 255), (0, 0, w - 30, 36), width=1, border_radius=6)
self.screen.blit(alert_s, (x + 15, y + h - 50))
alert_txt = self.font_alert.render(self.push_alert_text, True, COLOR_TEXT_PRIMARY)
txt_rect = alert_txt.get_rect(center=(x + w // 2, y + h - 32))
self.screen.blit(alert_txt, txt_rect)
# Model saved HUD notification popup
if self.save_alert_timer > 0:
alert_s = pygame.Surface((240, 32), pygame.SRCALPHA)
alert_s.fill((16, 185, 129, 220))
pygame.draw.rect(alert_s, (255, 255, 255), (0, 0, 240, 32), width=1, border_radius=6)
self.screen.blit(alert_s, (x + w - 255, y + 12))
save_txt = self.font_sub.render("[ MODEL CHECKPOINT SAVED ]", True, COLOR_TEXT_PRIMARY)
txt_rect = save_txt.get_rect(center=(x + w - 135, y + 28))
self.screen.blit(save_txt, txt_rect)
def _draw_glass_pill(self, x, y, w, h, label, val, color):
s = pygame.Surface((w, h), pygame.SRCALPHA)
s.fill((10, 14, 23, 220))
pygame.draw.rect(s, COLOR_PANEL_BORDER, (0, 0, w, h), width=1, border_radius=4)
self.screen.blit(s, (x, y))
lbl_s = self.font_small.render(label, True, COLOR_TEXT_MUTED)
val_s = self.font_sub.render(val, True, color)
self.screen.blit(lbl_s, (x + 8, y + 2))
self.screen.blit(val_s, (x + 8, y + 14))
def _draw_stat_cards(self, x, y, w, h):
"""Top-right Bento Grid Key Performance Indicator (KPI) cards."""
card_w = (w - 30) // 4
card_h = h
avg_20 = float(np.mean(list(self.recent_20_rewards))) if len(self.recent_20_rewards) > 0 else 0.0
cards_data = [
("EPISODE REWARD", f"{self.sim.episode_reward:.1f}", COLOR_CYAN),
("PEAK REWARD", f"{self.best_reward:.1f}", COLOR_AMBER),
("20-EP MOVING AVG", f"{avg_20:.1f}", COLOR_EMERALD),
("TOTAL TIMESTEPS", f"{self.trainer.total_timesteps:,}", COLOR_VIOLET),
]
for i, (lbl, val, col) in enumerate(cards_data):
cx = x + i * (card_w + 10)
rect = pygame.Rect(cx, y, card_w, card_h)
pygame.draw.rect(self.screen, COLOR_PANEL, rect, border_radius=8)
pygame.draw.rect(self.screen, COLOR_PANEL_BORDER, rect, width=1, border_radius=8)
lbl_surf = self.font_card_lbl.render(lbl, True, COLOR_TEXT_MUTED)
val_surf = self.font_card_val.render(val, True, col)
self.screen.blit(lbl_surf, (cx + 14, y + 14))
self.screen.blit(val_surf, (cx + 14, y + 42))
def _draw_learning_diagnostics(self, x, y, w, h):
"""PPO training diagnostic metrics card (Loss & Entropy)."""
rect = pygame.Rect(x, y, w, h)
pygame.draw.rect(self.screen, COLOR_PANEL, rect, border_radius=8)
pygame.draw.rect(self.screen, COLOR_PANEL_BORDER, rect, width=1, border_radius=8)
t_surf = self.font_card_lbl.render("LEARNING DIAGNOSTICS // PPO TELEMETRY", True, COLOR_TEXT_PRIMARY)
self.screen.blit(t_surf, (x + 14, y + 12))
pygame.draw.line(self.screen, COLOR_PANEL_BORDER, (x + 14, y + 30), (x + w - 14, y + 30), 1)
p_loss_str = f"Policy Gradient Loss : {self.trainer.policy_loss:+.4f}"
v_loss_str = f"Value Function Loss : {self.trainer.value_loss:.4f}"
ent_str = f"Entropy (Noise) : {self.trainer.entropy:.3f} | Iterations: #{self.trainer.update_count}"
self.screen.blit(self.font_mono.render(p_loss_str, True, COLOR_ROSE), (x + 18, y + 42))
self.screen.blit(self.font_mono.render(v_loss_str, True, COLOR_AMBER), (x + 18, y + 66))
self.screen.blit(self.font_mono.render(ent_str, True, COLOR_EMERALD), (x + 18, y + 90))
def _draw_actuator_torques(self, x, y, w, h):
"""Bottom-right 17-DOF joint actuator torque visualizers."""
rect = pygame.Rect(x, y, w, h)
pygame.draw.rect(self.screen, COLOR_PANEL, rect, border_radius=8)
pygame.draw.rect(self.screen, COLOR_PANEL_BORDER, rect, width=1, border_radius=8)
t_surf = self.font_card_lbl.render("ACTUATOR TORQUES // 17 DOF CONTINUOUS CONTROL", True, COLOR_TEXT_PRIMARY)
self.screen.blit(t_surf, (x + 14, y + 12))
pygame.draw.line(self.screen, COLOR_PANEL_BORDER, (x + 14, y + 30), (x + w - 14, y + 30), 1)
col_w = (w - 40) // 2
col1_x = x + 15
col2_x = x + col_w + 25
row_h = 36
start_y = y + 40
for i, name in enumerate(SimulationEngine.JOINT_NAMES):
val = self.sim.current_torques[i] if i < len(self.sim.current_torques) else 0.0
if i < 9:
bx = col1_x
by = start_y + i * row_h
else:
bx = col2_x
by = start_y + (i - 9) * row_h
name_surf = self.font_mono.render(name, True, COLOR_TEXT_SECONDARY)
self.screen.blit(name_surf, (bx, by))
bar_x = bx + 100
bar_w = col_w - 110
bar_h = 10
bar_rect = pygame.Rect(bar_x, by + 3, bar_w, bar_h)
pygame.draw.rect(self.screen, COLOR_PANEL_SUB, bar_rect, border_radius=3)
mid_x = bar_x + bar_w // 2
fill_w = int((abs(val)) * (bar_w // 2))
bar_col = COLOR_MAGENTA if val < 0 else COLOR_CYAN
if val < 0:
fill_rect = pygame.Rect(mid_x - fill_w, by + 3, fill_w, bar_h)
else:
fill_rect = pygame.Rect(mid_x, by + 3, fill_w, bar_h)
pygame.draw.rect(self.screen, bar_col, fill_rect, border_radius=2)
pygame.draw.line(self.screen, COLOR_PANEL_BORDER, (mid_x, by + 1), (mid_x, by + bar_h + 4), 1)
def _draw_reward_curve(self, x, y, w, h):
"""Bottom-left reward telemetry and moving average curves."""
rect = pygame.Rect(x, y, w, h)
pygame.draw.rect(self.screen, COLOR_PANEL, rect, border_radius=8)
pygame.draw.rect(self.screen, COLOR_PANEL_BORDER, rect, width=1, border_radius=8)
t_surf = self.font_card_lbl.render("REWARD TELEMETRY CURVE", True, COLOR_TEXT_PRIMARY)
self.screen.blit(t_surf, (x + 14, y + 12))
plot_x = x + 14
plot_y = y + 36
plot_w = w - 28
plot_h = h - 48
pygame.draw.rect(self.screen, COLOR_PANEL_SUB, (plot_x, plot_y, plot_w, plot_h), border_radius=4)
pygame.draw.line(self.screen, COLOR_PANEL_BORDER, (plot_x, plot_y + plot_h // 2), (plot_x + plot_w, plot_y + plot_h // 2), 1)
if len(self.recent_rewards) >= 2:
all_r = list(self.recent_rewards)
min_r = min(min(all_r), 0.0)
max_r = max(max(all_r), min_r + 50.0)
pad_y = 10
usable_h = plot_h - (pad_y * 2)
points = []
for i, r in enumerate(all_r):
px = plot_x + int(i * (plot_w / max(len(all_r) - 1, 1)))
norm_y = float(np.clip((r - min_r) / (max_r - min_r + 1e-5), 0.0, 1.0))
py = plot_y + plot_h - pad_y - int(norm_y * usable_h)
points.append((px, py))
pygame.draw.lines(self.screen, COLOR_CYAN, False, points, width=2)
pygame.draw.circle(self.screen, COLOR_AMBER, points[-1], 4)
max_lbl = self.font_mono.render(f"{int(max_r)}", True, COLOR_TEXT_MUTED)
min_lbl = self.font_mono.render(f"{int(min_r)}", True, COLOR_TEXT_MUTED)
self.screen.blit(max_lbl, (plot_x + plot_w - 32, plot_y + 4))
self.screen.blit(min_lbl, (plot_x + plot_w - 32, plot_y + plot_h - 16))
def _draw_console_logs(self, x, y, w, h):
"""Bottom-center system event logger."""
rect = pygame.Rect(x, y, w, h)
pygame.draw.rect(self.screen, COLOR_PANEL, rect, border_radius=8)
pygame.draw.rect(self.screen, COLOR_PANEL_BORDER, rect, width=1, border_radius=8)
t_surf = self.font_card_lbl.render("SYSTEM CONSOLE LOG", True, COLOR_TEXT_PRIMARY)
self.screen.blit(t_surf, (x + 14, y + 12))
log_box = pygame.Rect(x + 10, y + 34, w - 20, h - 44)
pygame.draw.rect(self.screen, COLOR_PANEL_SUB, log_box, border_radius=4)
for i, log in enumerate(self.console_logs):
is_latest = (i == len(self.console_logs) - 1)
col = COLOR_CYAN if is_latest else COLOR_TEXT_MUTED
txt_surf = self.font_mono.render(log, True, col)
self.screen.blit(txt_surf, (x + 16, y + 42 + i * 36))
def _draw_footer(self):
"""Bottom keyboard shortcuts navigation bar."""
footer_y = self.height - 36
guide_text = "[Space: Pause/Resume] [T: Toggle 3D] [F: Speed] [P: Push] [S: Save] [L: Load] [R/E: Reset]"
guide_surf = self.font_mono.render(guide_text, True, COLOR_TEXT_MUTED)
self.screen.blit(guide_surf, (20, footer_y))
if __name__ == "__main__":
app = RobotAILabGUI(width=1280, height=820)
app.run()
|