[Newton] Add initial design for Visualizers (#3979)

# Description

Initial design to support multiple visualizers with Isaac Lab 3.0

The visualizers
- OV Visualizer
- Newton OpenGL Visualizer
- Newton Rerun Visualizer

Each visualizer comes with a config class.

CLI newton_visualizer argument is removed.
- Discussion on how we can best enable visualizer with CLI arg is
ongoing.
- Currently, visualizers can be selected in the code by visualizer_cfgs
in simulation_cfg
  and the --headless arg disables any selected visualizer

Also there is a new abstraction called scene data provider which manages
per step data flow to visualizers.
This is a pretty thin abstraction right now and subject to change with
the renderers design and ov sdk.

---------

Signed-off-by: matthewtrepte <mtrepte@nvidia.com>
Co-authored-by: Antoine Richard <antoiner@nvidia.com>

Author: matthewtrepte

scripts/environments/random_agent.py

@@ -18,7 +18,6 @@
 )
 parser.add_argument("--num_envs", type=int, default=None, help="Number of environments to simulate.")
 parser.add_argument("--task", type=str, default=None, help="Name of the task.")
-parser.add_argument("--newton_visualizer", action="store_true", default=False, help="Enable Newton rendering.")
 # append AppLauncher cli args
 AppLauncher.add_app_launcher_args(parser)
 # parse the arguments
@@ -52,8 +51,8 @@ def main():
         device=args_cli.device,
         num_envs=args_cli.num_envs,
         use_fabric=not args_cli.disable_fabric,
-        newton_visualizer=args_cli.newton_visualizer,
     )
+
     # create environment
     env = gym.make(args_cli.task, cfg=env_cfg)
 

scripts/environments/zero_agent.py

@@ -18,7 +18,6 @@
 )
 parser.add_argument("--num_envs", type=int, default=None, help="Number of environments to simulate.")
 parser.add_argument("--task", type=str, default=None, help="Name of the task.")
-parser.add_argument("--newton_visualizer", action="store_true", default=False, help="Enable Newton rendering.")
 # append AppLauncher cli args
 AppLauncher.add_app_launcher_args(parser)
 # parse the arguments
@@ -52,8 +51,8 @@ def main():
         device=args_cli.device,
         num_envs=args_cli.num_envs,
         use_fabric=not args_cli.disable_fabric,
-        newton_visualizer=args_cli.newton_visualizer,
     )
+
     # create environment
     env = gym.make(args_cli.task, cfg=env_cfg)
 

scripts/reinforcement_learning/rl_games/play.py

@@ -32,7 +32,6 @@
     help="When no checkpoint provided, use the last saved model. Otherwise use the best saved model.",
 )
 parser.add_argument("--real-time", action="store_true", default=False, help="Run in real-time, if possible.")
-parser.add_argument("--newton_visualizer", action="store_true", default=False, help="Enable Newton rendering.")
 # append AppLauncher cli args
 AppLauncher.add_app_launcher_args(parser)
 # parse the arguments
@@ -80,12 +79,9 @@ def main():
     task_name = args_cli.task.split(":")[-1]
     # parse env configuration
     env_cfg = parse_env_cfg(
-        args_cli.task,
-        device=args_cli.device,
-        num_envs=args_cli.num_envs,
-        use_fabric=not args_cli.disable_fabric,
-        newton_visualizer=args_cli.newton_visualizer,
+        args_cli.task, device=args_cli.device, num_envs=args_cli.num_envs, use_fabric=not args_cli.disable_fabric
     )
+
     agent_cfg = load_cfg_from_registry(args_cli.task, "rl_games_cfg_entry_point")
 
     # specify directory for logging experiments

scripts/reinforcement_learning/rl_games/train.py

@@ -38,7 +38,6 @@
     const=True,
     help="if toggled, this experiment will be tracked with Weights and Biases",
 )
-parser.add_argument("--newton_visualizer", action="store_true", default=False, help="Enable Newton rendering.")
 # append AppLauncher cli args
 AppLauncher.add_app_launcher_args(parser)
 # parse the arguments
@@ -90,7 +89,6 @@ def main(env_cfg: ManagerBasedRLEnvCfg | DirectRLEnvCfg, agent_cfg: dict):
     # override configurations with non-hydra CLI arguments
     env_cfg.scene.num_envs = args_cli.num_envs if args_cli.num_envs is not None else env_cfg.scene.num_envs
     env_cfg.sim.device = args_cli.device if args_cli.device is not None else env_cfg.sim.device
-    env_cfg.sim.enable_newton_rendering = args_cli.newton_visualizer
 
     # randomly sample a seed if seed = -1
     if args_cli.seed == -1:

scripts/reinforcement_learning/rsl_rl/play.py

@@ -34,7 +34,6 @@
     help="Use the pre-trained checkpoint from Nucleus.",
 )
 parser.add_argument("--real-time", action="store_true", default=False, help="Run in real-time, if possible.")
-parser.add_argument("--newton_visualizer", action="store_true", default=False, help="Enable Newton rendering.")
 # append RSL-RL cli arguments
 cli_args.add_rsl_rl_args(parser)
 # append AppLauncher cli args
@@ -96,7 +95,6 @@ def main(env_cfg: ManagerBasedRLEnvCfg | DirectRLEnvCfg, agent_cfg: RslRlBaseRun
     # note: certain randomizations occur in the environment initialization so we set the seed here
     env_cfg.seed = agent_cfg.seed
     env_cfg.sim.device = args_cli.device if args_cli.device is not None else env_cfg.sim.device
-    env_cfg.sim.enable_newton_rendering = args_cli.newton_visualizer
 
     # specify directory for logging experiments
     log_root_path = os.path.join("logs", "rsl_rl", agent_cfg.experiment_name)
@@ -117,19 +115,6 @@ def main(env_cfg: ManagerBasedRLEnvCfg | DirectRLEnvCfg, agent_cfg: RslRlBaseRun
     # set the log directory for the environment (works for all environment types)
     env_cfg.log_dir = log_dir
 
-    # Set play mode for Newton viewer if using Newton visualizer
-    if args_cli.newton_visualizer:
-        # Set visualizer to play mode in Newton config
-        if hasattr(env_cfg.sim, "newton_cfg"):
-            env_cfg.sim.newton_cfg.visualizer_train_mode = False
-        else:
-            # Create newton_cfg if it doesn't exist
-            from isaaclab.sim._impl.newton_manager_cfg import NewtonCfg
-
-            newton_cfg = NewtonCfg()
-            newton_cfg.visualizer_train_mode = False
-            env_cfg.sim.newton_cfg = newton_cfg
-
     # create isaac environment
     env = gym.make(args_cli.task, cfg=env_cfg, render_mode="rgb_array" if args_cli.video else None)
 

scripts/reinforcement_learning/rsl_rl/train.py

@@ -32,7 +32,6 @@
     "--distributed", action="store_true", default=False, help="Run training with multiple GPUs or nodes."
 )
 parser.add_argument("--export_io_descriptors", action="store_true", default=False, help="Export IO descriptors.")
-parser.add_argument("--newton_visualizer", action="store_true", default=False, help="Enable Newton rendering.")
 # append RSL-RL cli arguments
 cli_args.add_rsl_rl_args(parser)
 # append AppLauncher cli args
@@ -122,7 +121,6 @@ def main(env_cfg: ManagerBasedRLEnvCfg | DirectRLEnvCfg, agent_cfg: RslRlBaseRun
     # note: certain randomizations occur in the environment initialization so we set the seed here
     env_cfg.seed = agent_cfg.seed
     env_cfg.sim.device = args_cli.device if args_cli.device is not None else env_cfg.sim.device
-    env_cfg.sim.enable_newton_rendering = args_cli.newton_visualizer
 
     # multi-gpu training configuration
     if args_cli.distributed:

scripts/reinforcement_learning/sb3/play.py

@@ -39,7 +39,6 @@
     default=False,
     help="Use a slower SB3 wrapper but keep all the extra training info.",
 )
-parser.add_argument("--newton_visualizer", action="store_true", default=False, help="Enable Newton rendering.")
 # append AppLauncher cli args
 AppLauncher.add_app_launcher_args(parser)
 # parse the arguments
@@ -83,11 +82,7 @@ def main():
     """Play with stable-baselines agent."""
     # parse configuration
     env_cfg = parse_env_cfg(
-        args_cli.task,
-        device=args_cli.device,
-        num_envs=args_cli.num_envs,
-        use_fabric=not args_cli.disable_fabric,
-        newton_visualizer=args_cli.newton_visualizer,
+        args_cli.task, device=args_cli.device, num_envs=args_cli.num_envs, use_fabric=not args_cli.disable_fabric
     )
 
     task_name = args_cli.task.split(":")[-1]

scripts/reinforcement_learning/sb3/train.py

@@ -32,7 +32,6 @@
     default=False,
     help="Use a slower SB3 wrapper but keep all the extra training info.",
 )
-parser.add_argument("--newton_visualizer", action="store_true", default=False, help="Enable Newton rendering.")
 # append AppLauncher cli args
 AppLauncher.add_app_launcher_args(parser)
 # parse the arguments
@@ -113,7 +112,6 @@ def main(env_cfg: ManagerBasedRLEnvCfg | DirectRLEnvCfg, agent_cfg: dict):
     # note: certain randomizations occur in the environment initialization so we set the seed here
     env_cfg.seed = agent_cfg["seed"]
     env_cfg.sim.device = args_cli.device if args_cli.device is not None else env_cfg.sim.device
-    env_cfg.sim.enable_newton_rendering = args_cli.newton_visualizer
 
     # directory for logging into
     run_info = datetime.now().strftime("%Y-%m-%d_%H-%M-%S")

scripts/reinforcement_learning/skrl/play.py

@@ -46,7 +46,6 @@
     help="The RL algorithm used for training the skrl agent.",
 )
 parser.add_argument("--real-time", action="store_true", default=False, help="Run in real-time, if possible.")
-parser.add_argument("--newton_visualizer", action="store_true", default=False, help="Enable Newton rendering.")
 
 # append AppLauncher cli args
 AppLauncher.add_app_launcher_args(parser)
@@ -112,11 +111,7 @@ def main():
 
     # parse configuration
     env_cfg = parse_env_cfg(
-        args_cli.task,
-        device=args_cli.device,
-        num_envs=args_cli.num_envs,
-        use_fabric=not args_cli.disable_fabric,
-        newton_visualizer=args_cli.newton_visualizer,
+        args_cli.task, device=args_cli.device, num_envs=args_cli.num_envs, use_fabric=not args_cli.disable_fabric
     )
     try:
         experiment_cfg = load_cfg_from_registry(task_name, f"skrl_{algorithm}_cfg_entry_point")

scripts/reinforcement_learning/skrl/train.py

@@ -44,7 +44,6 @@
     choices=["AMP", "PPO", "IPPO", "MAPPO"],
     help="The RL algorithm used for training the skrl agent.",
 )
-parser.add_argument("--newton_visualizer", action="store_true", default=False, help="Enable Newton rendering.")
 
 # append AppLauncher cli args
 AppLauncher.add_app_launcher_args(parser)
@@ -113,7 +112,6 @@ def main(env_cfg: ManagerBasedRLEnvCfg | DirectRLEnvCfg, agent_cfg: dict):
     # override configurations with non-hydra CLI arguments
     env_cfg.scene.num_envs = args_cli.num_envs if args_cli.num_envs is not None else env_cfg.scene.num_envs
     env_cfg.sim.device = args_cli.device if args_cli.device is not None else env_cfg.sim.device
-    env_cfg.sim.enable_newton_rendering = args_cli.newton_visualizer
 
     # multi-gpu training config
     if args_cli.distributed: