基于 IsaacLab 从零训练机器人行走

Python 环境与 Isaac Lab 安装

推荐使用 Anaconda 管理环境。创建虚拟环境并安装 PyTorch（CUDA 12.4）：

conda create -n env_isaaclab python=3.10
conda activate env_isaaclab
pip install --upgrade pip
conda install pytorch==2.4.0 torchvision==0.19.0 torchaudio==2.4.0 pytorch-cuda=12.4-c pytorch -c nvidia

接着安装 Isaac Sim 及 Isaac Lab。Isaac Sim 首次运行需接受 EULA 协议。

pip install 'isaacsim[all,extscache]==4.5.0' --extra-index-url https://pypi.nvidia.com
isaacsim

资产包可从官网下载（约 92GB），替换默认路径下的资产目录。最后克隆 Isaac Lab 代码库并安装依赖：

git clone [email protected]:isaac-sim/IsaacLab.git
cd IsaacLab
./isaaclab.sh --install
# 验证安装
./isaaclab.sh -p scripts/tutorials/00_sim/create_empty.py

机器人注册

获取 USD 文件

将 SolidWorks 导出的 URDF 文件导入 Isaac Sim，生成 USD 文件。注意检查 STL 路径格式，若使用 package://... 不可行，建议替换为相对或绝对路径。移动机器人需勾选可动基座。

配置文件编写

进入 Isaac Lab 任务配置目录，复制现有机器人配置文件夹并重命名（例如 op3）。

1. 机器人参数 (op3.py)

定义机器人的初始状态、关节名称及执行器参数。注意替换为你的实际关节名。

import isaaclab.sim as sim_utils
from isaaclab.actuators import ImplicitActuatorCfg
from isaaclab.assets.articulation import ArticulationCfg
from isaaclab.utils.assets import ISAAC_NUCLEUS_DIR

OP3_CFG = ArticulationCfg(
    spawn=sim_utils.UsdFileCfg(
        usd_path="xxx/op3.usd",  # 替换为你前面导出的机器人 usd 文件路径
        activate_contact_sensors=True,
        rigid_props=sim_utils.RigidBodyPropertiesCfg(
            disable_gravity=False,
            retain_accelerations=False,
            linear_damping=0.0,
            angular_damping=0.0,
            max_linear_velocity=20.0,
            max_angular_velocity=20.0,
            max_depenetration_velocity=10.0,
        ),
        articulation_props=sim_utils.ArticulationRootPropertiesCfg(
            enabled_self_collisions=True,
            solver_position_iteration_count=4,
            solver_velocity_iteration_count=4,
        ),
    ),
    init_state=ArticulationCfg.InitialStateCfg(
        pos=(0.0, 0.0, 0.28),  # 机器人初始位置
        joint_pos={
            "head_pan": 0.0,
            "head_tilt": 0.0,
            ".*_hip_yaw": 0.0,
            ".*_hip_roll": 0.0,
            ".*_hip_pitch": 0.0,
            ".*_knee": 0.0,
            ".*_ank_pitch": 0.0,
            ".*_ank_roll": 0.0,
            ".*_sho_pitch": 0.0,
            ".*_sho_roll": 0.0,
            ".*_el": 0.0,
        },
        joint_vel={".*": 0.0},
    ),
    soft_joint_pos_limit_factor=0.7,
    actuators={
        "legs": ImplicitActuatorCfg(
            joint_names_expr=[".*_hip_yaw", ".*_hip_roll", ".*_hip_pitch", ".*_knee", ".*_ank_pitch", ".*_ank_roll"],
            effort_limit_sim=20,
            velocity_limit_sim=20.0,
            stiffness={".*_hip_yaw": 100.0, ".*_hip_roll": 100.0, ".*_hip_pitch": 100.0, ".*_knee": 100.0, ".*_ank_pitch": 100.0, ".*_ank_roll": 100.0,},
            damping={".*_hip_yaw": 20.0, ".*_hip_roll": 20.0, ".*_hip_pitch": 20.0, ".*_knee": 20.0, ".*_ank_pitch": 20.0, ".*_ank_roll": 20.0,},
        ),
        "arms": ImplicitActuatorCfg(
            joint_names_expr=[".*_sho_pitch", ".*_sho_roll", ".*_el"],
            effort_limit_sim=20,
            velocity_limit_sim=20.0,
            stiffness={".*_sho_pitch": 100.0, ".*_sho_roll": 100.0, ".*_el": 100.0,},
            damping={".*_sho_pitch": 20.0, ".*_sho_roll": 20.0, ".*_el": 20.0,},
        ),
        "head": ImplicitActuatorCfg(
            joint_names_expr=["head_pan", "head_tilt"],
            effort_limit_sim=20,
            velocity_limit_sim=20.0,
            stiffness={"head_pan": 100.0, "head_tilt": 100.0,},
            damping={"head_pan": 20.0, "head_tilt": 20.0,},
        ),
    },
)

2. 高度检查脚本 (check_op3.py)

用于确认机器人落地后的实际基座高度，以便调整初始位置。

import argparse
import torch
from isaaclab.app import AppLauncher
from isaaclab.assets import Articulation
from isaaclab.sim import SimulationContext
from op3 import OP3_CFG

def design_scene(sim):
    cfg = sim_utils.GroundPlaneCfg()
    cfg.func("/World/defaultGroundPlane", cfg)
    cfg = sim_utils.DomeLightCfg(intensity=2000.0, color=(0.75, 0.75, 0.75))
    cfg.func("/World/Light", cfg)
    origins = torch.tensor([[0.0, -1.0, 0.0], [0.0, 0.0, 0.0], [0.0, 1.0, 0.0]]).to(device=sim.device)
    hr2 = Articulation(OP3_CFG.replace(prim_path="/World/G1"))
    robots = [hr2]
    return robots, origins

def run_simulator(sim, robots, origins):
    sim_dt = sim.get_physics_dt()
    count = 0
    while simulation_app.is_running():
        if count % 1000 == 0:
            count = 0
        for index, robot in enumerate(robots):
            joint_pos, joint_vel = robot.data.default_joint_pos, robot.data.default_joint_vel
            robot.write_joint_state_to_sim(joint_pos, joint_vel)
            root_state = robot.data.default_root_state.clone()
            root_state[:, :3] += origins[index]
            robot.write_root_pose_to_sim(root_state[:, :7])
            robot.write_root_velocity_to_sim(root_state[:, 7:])
            robot.reset()
        for robot in robots:
            robot.set_joint_position_target(robot.data.default_joint_pos.clone())
            robot.write_data_to_sim()
            sim.step()
            count += 1
        for robot in robots:
            robot.update(sim_dt)
            root_pos = robot.data.root_pos_w
            print(f"Robot height (z): {root_pos[:, 2].item():.3f}")

if __name__ == "__main__":
    main()

3. 速度环境配置 (op3_velocity_env_cfg.py)

这是核心配置文件，定义了观测、动作、奖励及终止条件。主要修改点包括替换机器人名称、关节链接名及物理材质参数。

# Copyright (c) 2022-2025, The Isaac Lab Project Developers.
# SPDX-License-Identifier: BSD-3-Clause
import math
from dataclasses import MISSING
import isaaclab.sim as sim_utils
from isaaclab.assets import ArticulationCfg, AssetBaseCfg
from isaaclab.envs import ManagerBasedRLEnvCfg
from isaaclab.managers import CurriculumTermCfg as CurrTerm
from isaaclab.managers import EventTermCfg as EventTerm
from isaaclab.managers import ObservationGroupCfg as ObsGroup
from isaaclab.managers import ObservationTermCfg as ObsTerm
from isaaclab.managers import RewardTermCfg as RewTerm
from isaaclab.managers import SceneEntityCfg
from isaaclab.managers import TerminationTermCfg as DoneTerm
from isaaclab.scene import InteractiveSceneCfg
from isaaclab.sensors import ContactSensorCfg, RayCasterCfg, patterns
from isaaclab.terrains import TerrainImporterCfg
from isaaclab.utils import configclass
from isaaclab.utils.assets import ISAAC_NUCLEUS_DIR, ISAACLAB_NUCLEUS_DIR
from isaaclab.utils.noise import AdditiveUniformNoiseCfg as Unoise
import isaaclab_tasks.manager_based.locomotion.velocity.mdp as mdp
from isaaclab.terrains.config.rough import ROUGH_TERRAINS_CFG
from isaaclab.managers import ActionTermCfg

@configclass
class MySceneCfg(InteractiveSceneCfg):
    terrain = TerrainImporterCfg(
        prim_path="/World/ground",
        terrain_type="generator",
        terrain_generator=ROUGH_TERRAINS_CFG,
        max_init_terrain_level=5,
        collision_group=-1,
        physics_material=sim_utils.RigidBodyMaterialCfg(
            friction_combine_mode="multiply",
            restitution_combine_mode="multiply",
            static_friction=1.0,
            dynamic_friction=1.0,
        ),
        visual_material=sim_utils.MdlFileCfg(
            mdl_path=f"{ISAACLAB_NUCLEUS_DIR}/Materials/TilesMarbleSpiderWhiteBrickBondHoned/TilesMarbleSpiderWhiteBrickBondHoned.mdl",
            project_uvw=True,
            texture_scale=(0.25, 0.25),
        ),
        debug_vis=False,
    )
    robot: ArticulationCfg = MISSING
    height_scanner = RayCasterCfg(
        prim_path="{ENV_REGEX_NS}/Robot/robotis_op3/base",
        offset=RayCasterCfg.OffsetCfg(pos=(0.0, 0.0, 20.0)),
        attach_yaw_only=True,
        pattern_cfg=patterns.GridPatternCfg(resolution=0.1, size=[1.6, 1.0]),
        debug_vis=False,
        mesh_prim_paths=["/World/ground"],
    )
    contact_forces = ContactSensorCfg(prim_path="{ENV_REGEX_NS}/Robot/robotis_op3/.*", history_length=3, track_air_time=True)
    sky_light = AssetBaseCfg(
        prim_path="/World/skyLight",
        spawn=sim_utils.DomeLightCfg(
            intensity=750.0,
            texture_file=f"{ISAAC_NUCLEUS_DIR}/Materials/Textures/Skies/PolyHaven/kloofendal_43d_clear_puresky_4k.hdr",
        ),
    )

@configclass
class CommandsCfg:
    base_velocity = mdp.UniformVelocityCommandCfg(
        asset_name="robot",
        resampling_time_range=(10.0, 10.0),
        rel_standing_envs=0.02,
        rel_heading_envs=1.0,
        heading_command=True,
        heading_control_stiffness=0.5,
        debug_vis=True,
        ranges=mdp.UniformVelocityCommandCfg.Ranges(
            lin_vel_x=(-1.0, 1.0),
            lin_vel_y=(-1.0, 1.0),
            ang_vel_z=(-1.0, 1.0),
            heading=(-math.pi, math.pi),
        ),
    )

@configclass
class ActionsCfg:
    joint_pos = mdp.JointPositionActionCfg(asset_name="robot", joint_names=[".*"], scale=0.5, use_default_offset=True)

@configclass
class ObservationsCfg:
    @configclass
    class PolicyCfg(ObsGroup):
        base_lin_vel = ObsTerm(func=mdp.base_lin_vel, noise=Unoise(n_min=-0.1, n_max=0.1))
        base_ang_vel = ObsTerm(func=mdp.base_ang_vel, noise=Unoise(n_min=-0.2, n_max=0.2))
        projected_gravity = ObsTerm(func=mdp.projected_gravity, noise=Unoise(n_min=-0.05, n_max=0.05))
        velocity_commands = ObsTerm(func=mdp.generated_commands, params={"command_name": "base_velocity"})
        joint_pos = ObsTerm(func=mdp.joint_pos_rel, noise=Unoise(n_min=-0.01, n_max=0.01))
        joint_vel = ObsTerm(func=mdp.joint_vel_rel, noise=Unoise(n_min=-1.5, n_max=1.5))
        actions = ObsTerm(func=mdp.last_action)
        height_scan = ObsTerm(
            func=mdp.height_scan,
            params={"sensor_cfg": SceneEntityCfg("height_scanner")},
            noise=Unoise(n_min=-0.1, n_max=0.1),
            clip=(-1.0, 1.0),
        )
        def __post_init__(self):
            self.enable_corruption = True
            self.concatenate_terms = True
        policy: PolicyCfg = PolicyCfg()

@configclass
class EventCfg:
    physics_material = EventTerm(
        func=mdp.randomize_rigid_body_material,
        mode="startup",
        params={"asset_cfg": SceneEntityCfg("robot", body_names=".*"), "static_friction_range": (0.8, 0.8), "dynamic_friction_range": (0.6, 0.6), "restitution_range": (0.0, 0.0), "num_buckets": 64,},
    )
    add_base_mass = EventTerm(
        func=mdp.randomize_rigid_body_mass,
        mode="startup",
        params={"asset_cfg": SceneEntityCfg("robot", body_names="base"), "mass_distribution_params": (-5.0, 5.0), "operation": "add",},
    )
    reset_base = EventTerm(
        func=mdp.reset_root_state_uniform,
        mode="reset",
        params={"pose_range": {"x": (-0.5, 0.5), "y": (-0.5, 0.5), "yaw": (-3.14, 3.14)}, "velocity_range": {"x": (-0.5, 0.5), "y": (-0.5, 0.5), "z": (-0.5, 0.5), "roll": (-0.5, 0.5), "pitch": (-0.5, 0.5), "yaw": (-0.5, 0.5),},},
    )
    push_robot = EventTerm(
        func=mdp.push_by_setting_velocity,
        mode="interval",
        interval_range_s=(10.0, 15.0),
        params={"velocity_range": {"x": (-0.5, 0.5), "y": (-0.5, 0.5)}},
    )

@configclass
class RewardsCfg:
    track_lin_vel_xy_exp = RewTerm(func=mdp.track_lin_vel_xy_exp, weight=1.0, params={"command_name": "base_velocity", "std": math.sqrt(0.25)})
    track_ang_vel_z_exp = RewTerm(func=mdp.track_ang_vel_z_exp, weight=0.5, params={"command_name": "base_velocity", "std": math.sqrt(0.25)})
    lin_vel_z_l2 = RewTerm(func=mdp.lin_vel_z_l2, weight=-2.0)
    dof_torques_l2 = RewTerm(func=mdp.joint_torques_l2, weight=-1.0e-5)
    feet_air_time = RewTerm(func=mdp.feet_air_time, weight=0.125, params={"sensor_cfg": SceneEntityCfg("contact_forces", body_names=".*ank_roll_link"), "command_name": "base_velocity", "threshold": 0.5,})

@configclass
class TerminationsCfg:
    time_out = DoneTerm(func=mdp.time_out, time_out=True)
    base_contact = DoneTerm(func=mdp.illegal_contact, params={"sensor_cfg": SceneEntityCfg("contact_forces", body_names="base"), "threshold": 1.0,})

@configclass
class LocomotionVelocityRoughEnvCfg(ManagerBasedRLEnvCfg):
    scene: MySceneCfg = MySceneCfg(num_envs=4096, env_spacing=2.5)
    observations: ObservationsCfg = ObservationsCfg()
    actions: ActionsCfg = ActionsCfg()
    commands: CommandsCfg = CommandsCfg()
    rewards: RewardsCfg = RewardsCfg()
    terminations: TerminationsCfg = TerminationsCfg()
    events: EventCfg = EventCfg()
    curriculum: CurriculumCfg = CurriculumCfg()

    def __post_init__(self):
        self.decimation = 4
        self.episode_length_s = 20.0
        self.sim.dt = 0.005
        self.sim.render_interval = self.decimation
        self.sim.physics_material = self.scene.terrain.physics_material

4. 复杂地面环境 (rough_env_cfg.py)

针对崎岖地形优化奖励函数，鼓励步态交替并惩罚脚部滑动。

# ... imports ...
from .op3_velocity_env_cfg import LocomotionVelocityRoughEnvCfg
from op3 import OP3_CFG

@configclass
class OP3Rewards(RewardsCfg):
    termination_penalty = RewTerm(func=mdp.is_terminated, weight=-200.0)
    track_lin_vel_xy_exp = RewTerm(func=mdp.track_lin_vel_xy_yaw_frame_exp, weight=1.0, params={"command_name": "base_velocity", "std": 0.5})
    track_ang_vel_z_exp = RewTerm(func=mdp.track_ang_vel_z_world_exp, weight=1.0, params={"command_name": "base_velocity", "std": 0.5})
    feet_air_time = RewTerm(func=mdp.feet_air_time_positive_biped, weight=0.25, params={"command_name": "base_velocity", "sensor_cfg": SceneEntityCfg("contact_forces", body_names=".*ank_roll_link"), "threshold": 0.4,})
    feet_slide = RewTerm(func=mdp.feet_slide, weight=-0.25, params={"sensor_cfg": SceneEntityCfg("contact_forces", body_names=".*ank_roll_link"), "asset_cfg": SceneEntityCfg("robot", body_names=".*ank_roll_link"),})
    dof_pos_limits = RewTerm(func=mdp.joint_pos_limits, weight=-1.0, params={"asset_cfg": SceneEntityCfg("robot", joint_names=".*_ank_roll")})
    joint_deviation_hip = RewTerm(func=mdp.joint_deviation_l1, weight=-0.2, params={"asset_cfg": SceneEntityCfg("robot", joint_names=[".*_hip_roll", ".*_hip_pitch", ".*_hip_yaw"])},)

@configclass
class OP3RoughEnvCfg(LocomotionVelocityRoughEnvCfg):
    rewards: OP3Rewards = OP3Rewards()

    def __post_init__(self):
        super().__post_init__()
        self.scene.robot = OP3_CFG.replace(prim_path="{ENV_REGEX_NS}/Robot")
        if self.scene.height_scanner:
            self.scene.height_scanner.prim_path = "{ENV_REGEX_NS}/Robot/robotis_op3/base"
        self.events.push_robot = None
        self.events.add_base_mass = None
        self.events.reset_robot_joints.params["position_range"] = (1.0, 1.0)
        self.events.base_external_force_torque.params["asset_cfg"].body_names = [".*base"]
        self.terminations.base_contact.params["sensor_cfg"].body_names = ".*base"
        self.rewards.flat_orientation_l2.weight = -1.0
        self.commands.base_velocity.ranges.lin_vel_x = (0.5, 1.0)
        self.commands.base_velocity.ranges.lin_vel_y = (0.0, 0.0)

5. 平地环境 (flat_env_cfg.py)

简化地形配置，移除高度扫描和课程难度。

from .rough_env_cfg import OP3RoughEnvCfg

@configclass
class OP3FlatEnvCfg(OP3RoughEnvCfg):
    def __post_init__(self):
        super().__post_init__()
        self.scene.terrain.terrain_type = "plane"
        self.scene.terrain.terrain_generator = None
        self.scene.height_scanner = None
        self.observations.policy.height_scan = None
        self.curriculum.terrain_levels = None
        self.rewards.feet_air_time.weight = 1.0
        self.rewards.feet_air_time.params["threshold"] = 0.6

6. 环境注册 (__init__.py)

在 __init__.py 中注册 Gym 环境，区分训练模式（Train）和测试模式（Play）。

import gymnasium as gym
from . import agents

gym.register(id="Rough-OP3-train", entry_point="isaaclab.envs:ManagerBasedRLEnv", disable_env_checker=True, kwargs={"env_cfg_entry_point": f"{__name__}.rough_env_cfg:OP3RoughEnvCfg", "rsl_rl_cfg_entry_point": f"{agents.__name__}.rsl_rl_ppo_cfg:OP3RoughPPORunnerCfg", "skrl_cfg_entry_point": f"{agents.__name__}:skrl_rough_ppo_cfg.yaml",})
gym.register(id="Flat-OP3-train", entry_point="isaaclab.envs:ManagerBasedRLEnv", disable_env_checker=True, kwargs={"env_cfg_entry_point": f"{__name__}.flat_env_cfg:OP3FlatEnvCfg", "rsl_rl_cfg_entry_point": f"{agents.__name__}.rsl_rl_ppo_cfg:OP3FlatPPORunnerCfg", "skrl_cfg_entry_point": f"{agents.__name__}:skrl_flat_ppo_cfg.yaml",})

7. 训练参数 (rsl_rl_ppo_cfg.py)

配置 PPO 算法超参数，如学习率、网络结构等。

from isaaclab.utils import configclass
from isaaclab_rl.rsl_rl import RslRlOnPolicyRunnerCfg, RslRlPpoActorCriticCfg, RslRlPpoAlgorithmCfg

@configclass
class OP3RoughPPORunnerCfg(RslRlOnPolicyRunnerCfg):
    num_steps_per_env = 24
    max_iterations = 3000
    save_interval = 50
    experiment_name = "OP3_rough"
    empirical_normalization = False
    policy = RslRlPpoActorCriticCfg(init_noise_std=1.0, actor_hidden_dims=[512, 256, 128], critic_hidden_dims=[512, 256, 128], activation="elu")
    algorithm = RslRlPpoAlgorithmCfg(value_loss_coef=1.0, use_clipped_value_loss=True, clip_param=0.2, entropy_coef=0.01, learning_rate=1.0e-3, gamma=0.99, lam=0.95)

训练与测试

启动训练

完成所有配置后，进入 Isaac Lab 目录执行训练命令。这里以平坦地形训练为例：

./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/train.py --task Flat-OP3-train --max_iterations 3000 --seed 42 --headless

如需查看帮助参数，可运行 -h。支持多 GPU 分布式训练及 TensorBoard 日志记录。

模型测试

训练完成后，在 ~/isaaclab/logs/rsl_rl/OP3_flat 目录下会生成 .pt 权重文件。选取最后一次保存的模型进行回放测试：

./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/play.py --task Flat-OP3-Play --checkpoint /home/user/isaaclab/logs/rsl_rl/OP3_flat/2025-06-05_19-31-40/model_2999.pt

此时应能看到机器人在仿真环境中按照训练策略行走。若表现不佳，可调整奖励权重或增加训练迭代次数。