GUI stays frozen while PhysX simulation continues to advance

[fan106]

Describe the bug

When I run an interactive state-based rollout in Isaac Sim, the GUI viewport stays completely static for the whole episode. The robot,
clutter, and floor all look frozen, and an object that should fall under gravity never moves. However, the rollout logs show non-zero
joint and body-state deltas, so the simulation appears to be advancing and the issue looks like a render / sync problem rather than a
physics failure.

Environment

• Isaac Sim: 6.0.0.1
• Isaac Lab: 3.0.0 beta
• Mode: GUI / interactive
• Envs: 1

Symptoms

• The entire visible scene remains frozen
• Clutter that should drop does not move
• Joint/body traces continue to change
• The robot base/root stays fixed, which is expected for a fixed-base arm

Also observed

The robot is initialized with zero joint values and may be intersecting the floor at import, but that does not explain why every visible
object in the scene stays static.

Why this matters

This makes interactive debugging unusable, because the simulation state and the viewport no longer match.

Expected

The viewport should reflect live simulation updates during rollout, including robot articulation motion and object motion.

Questions

1. Is this a Fabric / Hydra / USD synchronization issue in GUI mode?
2. Is there a required render setting for interactive state-based evaluation?
3. Is there a supported pattern for keeping the viewport synced when one script selects between multiple robot variants?

terminal output:
here is no error, and in the terminal print, the robot run successfully

[eval-state-step] step=36 dones=0 terminated=[False] time_outs=[False] delta=physx_step=+1; viewport_frame=+1; robot_root_pos_w_delta=(+0.000000, +0.000000, +0.000000) l2=0.000000 max_abs=0.000000; robot_root_quat_w_delta=(+0.000000, +0.000000, +0.000000, +0.000000) l2=0.000000 max_abs=0.000000; robot_root_lin_vel_w_delta=(+0.000000, +0.000000, +0.000000) l2=0.000000 max_abs=0.000000; robot_root_ang_vel_w_delta=(+0.000000, +0.000000, +0.000000) l2=0.000000 max_abs=0.000000; arm_joint_pos_delta=(-0.021552, -0.042036, -0.010389, -0.036366, -0.023099, +0.009462, -0.016961) l2=0.067622 max_abs=0.042036; arm_joint_vel_delta=(+0.008541, -0.035040, +0.013733, -0.022146, +0.049188, +0.006244, -0.021076) l2=0.069874 max_abs=0.049188; target_pose_delta=(+0.000000, +0.000000, +0.000000, +0.000000, +0.000000, +0.000000, -0.000000) l2=0.000000 max_abs=0.000000; target_velocity_delta=(+0.000001, +0.000000, -0.000000, +0.000001, +0.000002, +0.000083) l2=0.000084 max_abs=0.000083; target_linear_velocity_delta=(+0.000001, +0.000000, -0.000000) l2=0.000001 max_abs=0.000001; target_angular_velocity_delta=(+0.000001, +0.000002, +0.000083) l2=0.000084 max_abs=0.000083; body_pos_w_delta=(+0.000000, +0.000000, +0.000000, -0.000000, -0.000000, -0.000000, -0.000000, -0.000000, ...) l2=0.077477 max_abs=0.024224; body_quat_w_delta=(+0.000000, +0.000000, +0.000000, +0.000000, +0.000000, +0.000000, -0.010741, -0.000868, ...) l2=0.391861 max_abs=0.147232; body_lin_vel_w_delta=(+0.000000, +0.000000, +0.000000, -0.000263, -0.000077, -0.000000, +0.000009, -0.000486, ...) l2=1.458091 max_abs=0.797514; body_ang_vel_w_delta=(+0.000000, +0.000000, +0.000000, -0.000000, -0.000000, +0.008541, -0.011160, -0.033671, ...) l2=5.108474 max_abs=2.987212; body_link_pose_w_delta=(+0.000000, +0.000000, +0.000000, +0.000000, +0.000000, +0.000000, +0.000000, -0.000000, ...) l2=0.399447 max_abs=0.147232; body_link_vel_w_delta=(+0.000000, +0.000000, +0.000000, +0.000000, +0.000000, +0.000000, +0.000000, +0.000002, ...) l2=5.311198 max_abs=2.987212; body_pose_w_delta=(+0.000000, +0.000000, +0.000000, +0.000000, +0.000000, +0.000000, +0.000000, -0.000000, ...) l2=0.399447 max_abs=0.147232; body_vel_w_delta=(+0.000000, +0.000000, +0.000000, +0.000000, +0.000000, +0.000000, -0.000263, -0.000077, ...) l2=5.312489 max_abs=2.987212; done[episode_length]: 35 -> 36
[eval-state-trace] step=36 raw_action=(-1.125929, -0.206708, -1.915561, -0.450277, -0.263487, -0.206319, +0.313129, -0.442902, -0.374139, +0.229790, +0.425605, -0.208778, +0.527244, -0.318513, -0.157224, -0.235026, +0.198811, +0.285553, +0.454924, +0.281570, +0.504102, -0.365141, +0.290980, +0.200866, -0.218555, +0.392980) action=(-1.000000, -0.206708, -1.000000, -0.450277, -0.263487, -0.206319, +0.313129, -0.442902, -0.374139, +0.229790, +0.425605, -0.208778, +0.527244, -0.318513, -0.157224, -0.235026, +0.198811, +0.285553, +0.454924, +0.281570, +0.504102, -0.365141, +0.290980, +0.200866, -0.218555, +0.392980) arm_action=(-1.125929, -0.206708, -1.915561, -0.450277, -0.263487, -0.206319) hand_action=(+0.313129, -0.442902, -0.374139, +0.229790, +0.425605, -0.208778, +0.527244, -0.318513, -0.157224, -0.235026, +0.198811, +0.285553, +0.454924, +0.281570, +0.504102, -0.365141, +0.290980, +0.200866, -0.218555, +0.392980)
[eval-state-trace] step=36 smoothed_action=(-1.000000, -0.235078, -1.000000, -0.449484, -0.300175, -0.193845, +0.294947, -0.438957, -0.359295, +0.197745, +0.426309, -0.214490, +0.524460, -0.325017, -0.155527, -0.233116, +0.184334, +0.273740, +0.450873, +0.280348, +0.495719, -0.359329, +0.295106, +0.202778, -0.215509, +0.387516) osc_arm_dof_movements=(-0.020980, -0.044332, -0.009410, -0.037533, -0.021627, +0.010073, -0.016721)
[eval-state-trace] step=36 joint_targets_arm=(-0.215143, -1.252097, -0.181314, +0.053650, -0.244979, +1.384608, -0.217891) joint_targets_hand=(+0.515391, -0.138700, -0.141077, -0.291692, +1.560400, -0.370000, +1.548500, -0.468300, -0.164400, -0.370000, -0.202044, +0.239998, +1.558500, +0.370000, +1.509671, -0.477700, +1.558500, +0.370000, -0.476800, +1.573500)
[eval-state-trace] step=36 link_pose_pos=(+0.147504, +0.317435, +0.624375) link_pose_quat=(+0.998088, +0.019555, -0.056035, +0.017269)
[eval-state-trace] step=36 target_pose=(+0.285473, +0.317905, +0.443021, -0.762554, -0.606109, -0.009897, -0.225931) target_velocity=(+0.000416, +0.000129, -0.000048, -0.000032, +0.004432, -0.000134) terminated=[False] time_outs=[False] done_causes={'success': tensor([False], device='cuda:0'), 'fallen': tensor([False], device='cuda:0'), 'invalid_state': tensor([False], device='cuda:0'), 'arm_contact': tensor([False], device='cuda:0'), 'arm_contact_force': tensor([0.], device='cuda:0'), 'arm_support_penetration': tensor([False], device='cuda:0'), 'hand_support_penetration': tensor([False], device='cuda:0'), 'arm_support_penetration_depth': tensor([0.], device='cuda:0'), 'hand_support_penetration_depth': tensor([0.], device='cuda:0'), 'terminated': tensor([False], device='cuda:0'), 'time_out': tensor([False], device='cuda:0'), 'episode_length': tensor([36], device='cuda:0'), 'target_object_id': tensor([17], device='cuda:0'), 'target_lift_height': tensor([0.], device='cuda:0'), 'target_success_height': tensor([0.5230], device='cuda:0'), 'fgp_fallen': tensor([False], device='cuda:0'), 'fgp_success': tensor([False], device='cuda:0'), 'planner_subgoal_arrived': tensor([False], device='cuda:0'), 'planner_subgoal_success': tensor([False], device='cuda:0'), 'planner_is_final': tensor([True], device='cuda:0'), 'final_target_object_id': tensor([17], device='cuda:0'), 'fgp_safety_reset': tensor([False], device='cuda:0'), 'fgp_safety_violation': tensor([False], device='cuda:0'), 'fgp_contact_force_reset': tensor([False], device='cuda:0'), 'fgp_contact_force_violation': tensor([False], device='cuda:0'), 'fgp_contact_force_bad_steps': tensor([0], device='cuda:0'), 'fgp_object_lin_vel_reset': tensor([False], device='cuda:0'), 'fgp_object_lin_vel_violation': tensor([False], device='cuda:0'), 'fgp_object_ang_vel_reset': tensor([False], device='cuda:0'), 'fgp_object_ang_vel_violation': tensor([False], device='cuda:0'), 'fgp_safety_done_enabled': tensor([True], device='cuda:0'), 'fgp_terminated': tensor([False], device='cuda:0'), 'clutter_settle_active': tensor([False], device='cuda:0'), 'clutter_invalid_state': tensor([False], device='cuda:0')}