Skip to content

Tune Newton MJWarp locomotion environments #6861

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
ooctipus wants to merge 8 commits into
base: develop
Choose a base branch
from
Open

Tune Newton MJWarp locomotion environments #6861

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
1113a64
fix locomotion
ooctipus Aug 3, 2026
9d881d2
Add locomotion tuning changelog fragments
ooctipus Aug 3, 2026
d832f83
tuned
ooctipus Aug 4, 2026
0c20987
tunned
ooctipus Aug 4, 2026
ec8d3ef
reverted velocity termination
ooctipus Aug 4, 2026
a446a82
remove extra args
ooctipus Aug 4, 2026
5df3020
remove export
ooctipus Aug 4, 2026
d96038b
remvoe extra export
ooctipus Aug 4, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions source/isaaclab/changelog.d/locomotion-newton-tuning.minor.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,5 @@
Added
^^^^^

* Added :func:`~isaaclab.envs.mdp.body_lin_vel_out_of_manual_limit` to terminate environments when an
articulation body exceeds a configured linear speed.
2 changes: 2 additions & 0 deletions source/isaaclab/isaaclab/envs/mdp/__init__.pyi
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -141,6 +141,7 @@ __all__ = [
"joint_pos_out_of_limit",
"joint_pos_out_of_manual_limit",
"joint_vel_out_of_limit",
"body_lin_vel_out_of_manual_limit",
"joint_vel_out_of_manual_limit",
"pose_command_success",
"root_height_below_minimum",
Expand Down Expand Up @@ -295,6 +296,7 @@ from .terminations import (
joint_pos_out_of_limit,
joint_pos_out_of_manual_limit,
joint_vel_out_of_limit,
body_lin_vel_out_of_manual_limit,
joint_vel_out_of_manual_limit,
pose_command_success,
root_height_below_minimum,
Expand Down
11 changes: 11 additions & 0 deletions source/isaaclab/isaaclab/envs/mdp/terminations.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -189,3 +189,14 @@ def illegal_contact(env: ManagerBasedRLEnv, threshold: float, sensor_cfg: SceneE
return torch.any(
torch.max(torch.linalg.norm(net_contact_forces[:, :, sensor_cfg.body_ids], dim=-1), dim=1)[0] > threshold, dim=1
)


def body_lin_vel_out_of_manual_limit(

@isaaclab-review-bot isaaclab-review-bot Bot Aug 3, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🟡 Warning · Implementation — Missing changelog fragments for touched packages

Repository rules require one fragment per touched package under source/<pkg>/changelog.d/. This PR touches isaaclab (new exported body_lin_vel_out_of_manual_limit) and isaaclab_tasks (physics preset retune, new body_speed termination, AnymalD init pose) but adds none, so these user-visible changes will be omitted from the compiled changelog. Add an Added fragment for the termination and a Changed fragment for the locomotion config updates.

env: ManagerBasedRLEnv, max_speed: float, asset_cfg: SceneEntityCfg = SceneEntityCfg("robot")
) -> torch.Tensor:
"""Terminate when any of the asset's bodies moves faster than the provided limit."""

@isaaclab-review-bot isaaclab-review-bot Bot Aug 3, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🔵 Suggestion · Api — Docstring lacks Args and SI unit

Public API docstrings must be Google-style with an Args: section and inline SI units for physical quantities. This newly exported termination documents neither max_speed ([m/s]) nor asset_cfg, so callers cannot tell how the threshold is interpreted. Add an Args: block using name: description form and annotate the speed unit.

# extract the used quantities (to enable type-hinting)
asset: Articulation = env.scene[asset_cfg.name]
# compute any violations
speed = torch.linalg.norm(asset.data.body_lin_vel_w.torch[:, asset_cfg.body_ids], dim=-1)
return torch.any(speed > max_speed, dim=1)
5 changes: 5 additions & 0 deletions source/isaaclab_tasks/changelog.d/locomotion-newton-tuning.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,5 @@
Fixed
^^^^^

* Fixed Newton MJWarp locomotion training instability by tuning contact capacity and parameters, adjusting the
ANYmal-D initial height, and terminating excessively fast bodies.
7 changes: 3 additions & 4 deletions source/isaaclab_tasks/isaaclab_tasks/core/velocity/config/anymal_d/rough_env_cfg.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,8 +19,7 @@ class AnymalDRoughEnvCfg(LocomotionVelocityRoughEnvCfg):
def __post_init__(self):
super().__post_init__()

# sim
# lower margin to avoid self-collision
self.sim.physics.newton_mjwarp.default_shape_cfg.margin = 0.001
# scene
self.scene.robot = ANYMAL_D_CFG.replace(prim_path="{ENV_REGEX_NS}/Robot")
self.scene.robot = ANYMAL_D_CFG.replace(
prim_path="{ENV_REGEX_NS}/Robot", init_state=ANYMAL_D_CFG.init_state.replace(pos=(0.0, 0.0, 0.65))
)
2 changes: 1 addition & 1 deletion source/isaaclab_tasks/isaaclab_tasks/core/velocity/config/cassie/rough_env_cfg.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,7 +57,7 @@ class CassieRoughEnvCfg(LocomotionVelocityRoughEnvCfg):

def __post_init__(self):
super().__post_init__()

self.sim.physics.newton_mjwarp.num_substeps = 2
# scene
self.scene.robot = CASSIE_CFG.replace(prim_path="{ENV_REGEX_NS}/Robot")
self.scene.robot.actuators["legs"].armature = preset(default=0.0, newton_mjwarp=0.02)
Expand Down
2 changes: 1 addition & 1 deletion source/isaaclab_tasks/isaaclab_tasks/core/velocity/config/g1/rough_env_cfg.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -110,7 +110,7 @@ class G1RoughEnvCfg(LocomotionVelocityRoughEnvCfg):

def __post_init__(self):
super().__post_init__()

self.sim.physics.newton_mjwarp.num_substeps = 2
# scene
self.scene.robot = G1_MINIMAL_CFG.replace(prim_path="{ENV_REGEX_NS}/Robot")
self.scene.height_scanner.prim_path = "{ENV_REGEX_NS}/Robot/torso_link"
Expand Down
12 changes: 6 additions & 6 deletions source/isaaclab_tasks/isaaclab_tasks/core/velocity/velocity_env_cfg.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,8 +52,8 @@ class RoughPhysicsCfg(PresetCfg):
physx = PhysxAutoCfg(isaacsim_physx=isaacsim_physx, ovphysx=ovphysx)
newton_mjwarp = NewtonCfg(
solver_cfg=MJWarpSolverCfg(
njmax=200,
nconmax=100,
njmax=1000,
nconmax=300,
cone="pyramidal",
impratio=1.0,
integrator="implicitfast",
Expand All @@ -62,10 +62,7 @@ class RoughPhysicsCfg(PresetCfg):
collision_cfg=NewtonCollisionPipelineCfg(max_triangle_pairs=2_500_000),
num_substeps=1,
debug_mode=False,
# 1 cm shape margin is the single most important Newton setting for rough
# terrain — without it, non-AnymalD robots fail to learn stable contact
# on triangle-mesh terrain. See isaaclab_newton 0.5.22 changelog.
default_shape_cfg=NewtonShapeCfg(margin=0.01),
default_shape_cfg=NewtonShapeCfg(margin=0.0, ke=160000.0, kd=1100.0),

@greptile-apps greptile-apps Bot Aug 3, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Zero margin breaks terrain contact

When inherited rough-terrain tasks run with the newton_mjwarp preset, this shared zero margin removes the nonzero margin required for stable triangle-mesh contact, causing affected robots to lose contact stability and fail to learn rough-terrain locomotion. It also combines with the removed AnymalD override to discard that robot's separately tuned margin.

Knowledge Base Used: isaaclab_tasks: Task Registration and Organization

@isaaclab-review-bot isaaclab-review-bot Bot Aug 3, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🟡 Warning · Design Architecture — Shared preset drops margin other robots rely on

RoughPhysicsCfg is inherited by every rough-terrain velocity task, not only AnymalD. The comment deleted from these exact lines stated the 1 cm margin was required for non-AnymalD robots to learn stable contact on triangle-mesh terrain, and the AnymalD-specific override was removed in the same change. Setting margin=0.0 (plus new ke/kd) in the shared preset therefore regresses the other inheriting configs. Keep the shared default and scope the retune to the AnymalD config.

)
default = isaacsim_physx

Expand Down Expand Up @@ -321,6 +318,7 @@ class TerminationsCfg:
func=mdp.illegal_contact,
params={"sensor_cfg": SceneEntityCfg("contact_forces", body_names="base"), "threshold": 1.0},
)
body_speed = DoneTerm(func=mdp.body_lin_vel_out_of_manual_limit, params={"max_speed": 20.0})


@configclass
Expand Down Expand Up @@ -362,6 +360,8 @@ def __post_init__(self):
self.sim.dt = 0.005
self.sim.render_interval = self.decimation
self.sim.physics_material = self.scene.terrain.physics_material
newton = self.sim.physics.newton_mjwarp
newton.collision_cfg.rigid_contact_max = newton.solver_cfg.nconmax * self.scene.num_envs

@greptile-apps greptile-apps Bot Aug 3, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Contact capacity becomes stale

rigid_contact_max is calculated from the construction-time environment count, but play mode and CLI/Hydra overrides change scene.num_envs afterward without recalculating it. This leaves play mode with a substantially oversized contact allocation and larger runs with a contact budget that no longer matches the requested number of environments.

Knowledge Base Used: isaaclab_tasks: Task Registration and Organization

@isaaclab-review-bot isaaclab-review-bot Bot Aug 3, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🟡 Warning · Implementation — Contact buffer sized from default num_envs

rigid_contact_max is computed in __post_init__, which runs at config construction while scene.num_envs is still the class default (4096). CLI/Hydra --num_envs overrides are applied to the instantiated config afterwards and do not re-run __post_init__, so the buffer stays at 300*4096 regardless: heavily over-allocated for small runs and undersized above 4096 envs. Derive this where num_envs is final.

# update sensor update periods
# we tick all the sensors based on the smallest update period (physics update period)
if self.scene.height_scanner is not None:
Expand Down
Loading