はじめに

Unitreeの4脚ロボットgo2の歩容取得はGenesisのexampleにあり、以下の記事でその手順が紹介されています。

今回は、デモのgo2をMITのMini Cheetahに変更して、学習して見ようと思います。

Mini Cheetah

MITのMini Cheetahは初めて4脚ロボットでバックフリップをしたロボットみたいです。

個人的に、オープンソースのMini CheetahのソフトウェアであるCheetah Softwareで4脚ロボットの勉強を始めたため、今回はMini Cheetahを選びました。

使用環境

ubuntu24.04
Python3.11
GPU:RTX3070

Genesisのセットアップに関しては以前記事を書いたので、そちらを参照してください。（宣伝）

学習

go2の学習のデモを少し変更し、Mini Cheetahの学習を行います。
go2の学習には強化学習環境のgo2_env.py、学習を行うgo2_train.py、動作確認を行うgo2_eval.pyの3つのファイルがあり、これらを変更します。

URDFの変更

Mini Cheetahのurdfには以下のリポジトリのものを使用します。

変更点として、mini_cheetah.urdfのmesh/~を../mesh/~に変更しました。

また、urdfが正しいかチェックするためにGenesisのURDFビュワー機能を使います。

gs view mini_cheetah.urdf

これでURDFが正しいことを確認できました。

参考ポスト

Genesis の URDFビューワー機能で、ロボットOSS「Meridian」で使用するヒューマノイドRoid1のURDFモデルを描画してみた

ROSの「rqt」みたくGUIで軸操作が可能

Tkinterのインストールが必要：
sudo apt-get install python3-tk
gs view <filename>#rosjp #Meridian計画 https://t.co/XDzdxrulx3 pic.twitter.com/schkRsAJMu
— holypong (@automo_emo) January 1, 2025

続いてgo2_env.pyをコピーして新しくmini_cheetah_env.pyを作ります。
以下は変更点になります。

gs.morphs.URDF(
                # file="urdf/go2/urdf/go2.urdf",
                file="Mini_Cheetah_test/mini_cheetah_urdf/urdf/mini_cheetah.urdf",
                pos=self.base_init_pos.cpu().numpy(),
                quat=self.base_init_quat.cpu().numpy(),
            ),

gs.morphs.URDFのfileをurdfのパスに変更してください。
（上記ではMini_Cheetah_test/mini_cheetah_urdf/urdf/にmini_cheetah.urdfを配置)

また、クラス名をGo2EnvからMiniCheetahEnvに変更しています。

以下はmini_cheetahの強化学習環境であるmini_cheetah_env.pyです。

mini_cheetah_env.py

import torch
import math
import genesis as gs
from genesis.utils.geom import quat_to_xyz, transform_by_quat, inv_quat, transform_quat_by_quat


def gs_rand_float(lower, upper, shape, device):
    return (upper - lower) * torch.rand(size=shape, device=device) + lower


class MiniCheetahEnv:
    def __init__(self, num_envs, env_cfg, obs_cfg, reward_cfg, command_cfg, show_viewer=False, device="cuda"):
        self.device = torch.device(device)

        self.num_envs = num_envs
        self.num_obs = obs_cfg["num_obs"]
        self.num_privileged_obs = None
        self.num_actions = env_cfg["num_actions"]
        self.num_commands = command_cfg["num_commands"]

        self.simulate_action_latency = True  # there is a 1 step latency on real robot
        self.dt = 0.02  # control frequence on real robot is 50hz
        self.max_episode_length = math.ceil(env_cfg["episode_length_s"] / self.dt)

        self.env_cfg = env_cfg
        self.obs_cfg = obs_cfg
        self.reward_cfg = reward_cfg
        self.command_cfg = command_cfg

        self.obs_scales = obs_cfg["obs_scales"]
        self.reward_scales = reward_cfg["reward_scales"]

        # create scene
        self.scene = gs.Scene(
            sim_options=gs.options.SimOptions(dt=self.dt, substeps=2),
            viewer_options=gs.options.ViewerOptions(
                max_FPS=int(0.5 / self.dt),
                camera_pos=(2.0, 0.0, 2.5),
                camera_lookat=(0.0, 0.0, 0.5),
                camera_fov=40,
            ),
            vis_options=gs.options.VisOptions(n_rendered_envs=1),
            rigid_options=gs.options.RigidOptions(
                dt=self.dt,
                constraint_solver=gs.constraint_solver.Newton,
                enable_collision=True,
                enable_joint_limit=True,
            ),
            show_viewer=show_viewer,
        )

        # add plain
        self.scene.add_entity(gs.morphs.URDF(file="urdf/plane/plane.urdf", fixed=True))

        # add robot
        self.base_init_pos = torch.tensor(self.env_cfg["base_init_pos"], device=self.device)
        self.base_init_quat = torch.tensor(self.env_cfg["base_init_quat"], device=self.device)
        self.inv_base_init_quat = inv_quat(self.base_init_quat)
        self.robot = self.scene.add_entity(
            gs.morphs.URDF(
                # file="urdf/go2/urdf/go2.urdf",
                file="Mini_Cheetah_test/mini_cheetah_urdf/urdf/mini_cheetah.urdf",
                pos=self.base_init_pos.cpu().numpy(),
                quat=self.base_init_quat.cpu().numpy(),
            ),
        )

        # build
        self.scene.build(n_envs=num_envs)

        # names to indices
        self.motor_dofs = [self.robot.get_joint(name).dof_idx_local for name in self.env_cfg["dof_names"]]

        # PD control parameters
        self.robot.set_dofs_kp([self.env_cfg["kp"]] * self.num_actions, self.motor_dofs)
        self.robot.set_dofs_kv([self.env_cfg["kd"]] * self.num_actions, self.motor_dofs)

        # prepare reward functions and multiply reward scales by dt
        self.reward_functions, self.episode_sums = dict(), dict()
        for name in self.reward_scales.keys():
            self.reward_scales[name] *= self.dt
            self.reward_functions[name] = getattr(self, "_reward_" + name)
            self.episode_sums[name] = torch.zeros((self.num_envs,), device=self.device, dtype=gs.tc_float)

        # initialize buffers
        self.base_lin_vel = torch.zeros((self.num_envs, 3), device=self.device, dtype=gs.tc_float)
        self.base_ang_vel = torch.zeros((self.num_envs, 3), device=self.device, dtype=gs.tc_float)
        self.projected_gravity = torch.zeros((self.num_envs, 3), device=self.device, dtype=gs.tc_float)
        self.global_gravity = torch.tensor([0.0, 0.0, -1.0], device=self.device, dtype=gs.tc_float).repeat(
            self.num_envs, 1
        )
        self.obs_buf = torch.zeros((self.num_envs, self.num_obs), device=self.device, dtype=gs.tc_float)
        self.rew_buf = torch.zeros((self.num_envs,), device=self.device, dtype=gs.tc_float)
        self.reset_buf = torch.ones((self.num_envs,), device=self.device, dtype=gs.tc_int)
        self.episode_length_buf = torch.zeros((self.num_envs,), device=self.device, dtype=gs.tc_int)
        self.commands = torch.zeros((self.num_envs, self.num_commands), device=self.device, dtype=gs.tc_float)
        self.commands_scale = torch.tensor(
            [self.obs_scales["lin_vel"], self.obs_scales["lin_vel"], self.obs_scales["ang_vel"]],
            device=self.device,
            dtype=gs.tc_float,
        )
        self.actions = torch.zeros((self.num_envs, self.num_actions), device=self.device, dtype=gs.tc_float)
        self.last_actions = torch.zeros_like(self.actions)
        self.dof_pos = torch.zeros_like(self.actions)
        self.dof_vel = torch.zeros_like(self.actions)
        self.last_dof_vel = torch.zeros_like(self.actions)
        self.base_pos = torch.zeros((self.num_envs, 3), device=self.device, dtype=gs.tc_float)
        self.base_quat = torch.zeros((self.num_envs, 4), device=self.device, dtype=gs.tc_float)
        self.default_dof_pos = torch.tensor(
            [self.env_cfg["default_joint_angles"][name] for name in self.env_cfg["dof_names"]],
            device=self.device,
            dtype=gs.tc_float,
        )
        self.extras = dict()  # extra information for logging

    def _resample_commands(self, envs_idx):
        self.commands[envs_idx, 0] = gs_rand_float(*self.command_cfg["lin_vel_x_range"], (len(envs_idx),), self.device)
        self.commands[envs_idx, 1] = gs_rand_float(*self.command_cfg["lin_vel_y_range"], (len(envs_idx),), self.device)
        self.commands[envs_idx, 2] = gs_rand_float(*self.command_cfg["ang_vel_range"], (len(envs_idx),), self.device)

    def step(self, actions):
        self.actions = torch.clip(actions, -self.env_cfg["clip_actions"], self.env_cfg["clip_actions"])
        exec_actions = self.last_actions if self.simulate_action_latency else self.actions
        target_dof_pos = exec_actions * self.env_cfg["action_scale"] + self.default_dof_pos
        self.robot.control_dofs_position(target_dof_pos, self.motor_dofs)
        self.scene.step()

        # update buffers
        self.episode_length_buf += 1
        self.base_pos[:] = self.robot.get_pos()
        self.base_quat[:] = self.robot.get_quat()
        self.base_euler = quat_to_xyz(
            transform_quat_by_quat(torch.ones_like(self.base_quat) * self.inv_base_init_quat, self.base_quat)
        )
        inv_base_quat = inv_quat(self.base_quat)
        self.base_lin_vel[:] = transform_by_quat(self.robot.get_vel(), inv_base_quat)
        self.base_ang_vel[:] = transform_by_quat(self.robot.get_ang(), inv_base_quat)
        self.projected_gravity = transform_by_quat(self.global_gravity, inv_base_quat)
        self.dof_pos[:] = self.robot.get_dofs_position(self.motor_dofs)
        self.dof_vel[:] = self.robot.get_dofs_velocity(self.motor_dofs)

        # resample commands
        envs_idx = (
            (self.episode_length_buf % int(self.env_cfg["resampling_time_s"] / self.dt) == 0)
            .nonzero(as_tuple=False)
            .flatten()
        )
        self._resample_commands(envs_idx)

        # check termination and reset
        self.reset_buf = self.episode_length_buf > self.max_episode_length
        self.reset_buf |= torch.abs(self.base_euler[:, 1]) > self.env_cfg["termination_if_pitch_greater_than"]
        self.reset_buf |= torch.abs(self.base_euler[:, 0]) > self.env_cfg["termination_if_roll_greater_than"]

        time_out_idx = (self.episode_length_buf > self.max_episode_length).nonzero(as_tuple=False).flatten()
        self.extras["time_outs"] = torch.zeros_like(self.reset_buf, device=self.device, dtype=gs.tc_float)
        self.extras["time_outs"][time_out_idx] = 1.0

        self.reset_idx(self.reset_buf.nonzero(as_tuple=False).flatten())

        # compute reward
        self.rew_buf[:] = 0.0
        for name, reward_func in self.reward_functions.items():
            rew = reward_func() * self.reward_scales[name]
            self.rew_buf += rew
            self.episode_sums[name] += rew

        # compute observations
        self.obs_buf = torch.cat(
            [
                self.base_ang_vel * self.obs_scales["ang_vel"],  # 3
                self.projected_gravity,  # 3
                self.commands * self.commands_scale,  # 3
                (self.dof_pos - self.default_dof_pos) * self.obs_scales["dof_pos"],  # 12
                self.dof_vel * self.obs_scales["dof_vel"],  # 12
                self.actions,  # 12
            ],
            axis=-1,
        )

        self.last_actions[:] = self.actions[:]
        self.last_dof_vel[:] = self.dof_vel[:]

        return self.obs_buf, None, self.rew_buf, self.reset_buf, self.extras

    def get_observations(self):
        return self.obs_buf

    def get_privileged_observations(self):
        return None

    def reset_idx(self, envs_idx):
        if len(envs_idx) == 0:
            return

        # reset dofs
        self.dof_pos[envs_idx] = self.default_dof_pos
        self.dof_vel[envs_idx] = 0.0
        self.robot.set_dofs_position(
            position=self.dof_pos[envs_idx],
            dofs_idx_local=self.motor_dofs,
            zero_velocity=True,
            envs_idx=envs_idx,
        )

        # reset base
        self.base_pos[envs_idx] = self.base_init_pos
        self.base_quat[envs_idx] = self.base_init_quat.reshape(1, -1)
        self.robot.set_pos(self.base_pos[envs_idx], zero_velocity=False, envs_idx=envs_idx)
        self.robot.set_quat(self.base_quat[envs_idx], zero_velocity=False, envs_idx=envs_idx)
        self.base_lin_vel[envs_idx] = 0
        self.base_ang_vel[envs_idx] = 0
        self.robot.zero_all_dofs_velocity(envs_idx)

        # reset buffers
        self.last_actions[envs_idx] = 0.0
        self.last_dof_vel[envs_idx] = 0.0
        self.episode_length_buf[envs_idx] = 0
        self.reset_buf[envs_idx] = True

        # fill extras
        self.extras["episode"] = {}
        for key in self.episode_sums.keys():
            self.extras["episode"]["rew_" + key] = (
                torch.mean(self.episode_sums[key][envs_idx]).item() / self.env_cfg["episode_length_s"]
            )
            self.episode_sums[key][envs_idx] = 0.0

        self._resample_commands(envs_idx)

    def reset(self):
        self.reset_buf[:] = True
        self.reset_idx(torch.arange(self.num_envs, device=self.device))
        return self.obs_buf, None

    # ------------ reward functions----------------
    def _reward_tracking_lin_vel(self):
        # Tracking of linear velocity commands (xy axes)
        lin_vel_error = torch.sum(torch.square(self.commands[:, :2] - self.base_lin_vel[:, :2]), dim=1)
        return torch.exp(-lin_vel_error / self.reward_cfg["tracking_sigma"])

    def _reward_tracking_ang_vel(self):
        # Tracking of angular velocity commands (yaw)
        ang_vel_error = torch.square(self.commands[:, 2] - self.base_ang_vel[:, 2])
        return torch.exp(-ang_vel_error / self.reward_cfg["tracking_sigma"])

    def _reward_lin_vel_z(self):
        # Penalize z axis base linear velocity
        return torch.square(self.base_lin_vel[:, 2])

    def _reward_action_rate(self):
        # Penalize changes in actions
        return torch.sum(torch.square(self.last_actions - self.actions), dim=1)

    def _reward_similar_to_default(self):
        # Penalize joint poses far away from default pose
        return torch.sum(torch.abs(self.dof_pos - self.default_dof_pos), dim=1)

    def _reward_base_height(self):
        # Penalize base height away from target
        return torch.square(self.base_pos[:, 2] - self.reward_cfg["base_height_target"])

ジョイント名の変更

上記のmini_cheetahのurdfのジョイント名は、go2のジョイント名と異なるため、それらを修正します。
go2_train.pyをコピーし、学習を行うためのmini_cheetah_train.pyを作成します。

go2とmini_cheetahのジョイント名の対応は以下のようになります。

go2	mini_cheetah
FL_hip_joint	torso_to_abduct_fl_j
FR_hip_joint	torso_to_abduct_fr_j
RL_hip_joint	torso_to_abduct_hr_j
RR_hip_joint	torso_to_abduct_hl_j
FL_thigh_joint	abduct_fl_to_thigh_fl_j
FR_thigh_joint	abduct_fr_to_thigh_fr_j
RL_thigh_joint	abduct_hr_to_thigh_hr_j
RR_thigh_joint	abduct_hl_to_thigh_hl_j
FL_calf_joint	thigh_fl_to_knee_fl_j
FR_calf_joint	thigh_fr_to_knee_fr_j
RL_calf_joint	thigh_hr_to_knee_hr_j
RR_calf_joint	thigh_hl_to_knee_hl_j

これを考慮しmini_cheetah_train.pyのget_cfgs()のdefault_joint_anglesとdof_nameを変更します。

また、先程クラス名を変更したので、imoportのモジュール名も変更しています。
変更後の学習用のmini_cheetah_train.pyは以下になります。

mini_cheetah_train.py

import argparse
import os
import pickle
import shutil

from mini_cheetah_env import MiniCheetahEnv
from rsl_rl.runners import OnPolicyRunner

import genesis as gs


def get_train_cfg(exp_name, max_iterations):

    train_cfg_dict = {
        "algorithm": {
            "clip_param": 0.2,
            "desired_kl": 0.01,
            "entropy_coef": 0.01,
            "gamma": 0.99,
            "lam": 0.95,
            "learning_rate": 0.001,
            "max_grad_norm": 1.0,
            "num_learning_epochs": 5,
            "num_mini_batches": 4,
            "schedule": "adaptive",
            "use_clipped_value_loss": True,
            "value_loss_coef": 1.0,
        },
        "init_member_classes": {},
        "policy": {
            "activation": "elu",
            "actor_hidden_dims": [512, 256, 128],
            "critic_hidden_dims": [512, 256, 128],
            "init_noise_std": 1.0,
        },
        "runner": {
            "algorithm_class_name": "PPO",
            "checkpoint": -1,
            "experiment_name": exp_name,
            "load_run": -1,
            "log_interval": 1,
            "max_iterations": max_iterations,
            "num_steps_per_env": 24,
            "policy_class_name": "ActorCritic",
            "record_interval": -1,
            "resume": False,
            "resume_path": None,
            "run_name": "",
            "runner_class_name": "runner_class_name",
            "save_interval": 100,
        },
        "runner_class_name": "OnPolicyRunner",
        "seed": 1,
    }

    return train_cfg_dict


def get_cfgs():
    env_cfg = {
        "num_actions": 12,
        # joint/link names
        # "default_joint_angles": {  # [rad]
        #     "FL_hip_joint": 0.0,
        #     "FR_hip_joint": 0.0,
        #     "RL_hip_joint": 0.0,
        #     "RR_hip_joint": 0.0,
        #     "FL_thigh_joint": 0.8,
        #     "FR_thigh_joint": 0.8,
        #     "RL_thigh_joint": 1.0,
        #     "RR_thigh_joint": 1.0,
        #     "FL_calf_joint": -1.5,
        #     "FR_calf_joint": -1.5,
        #     "RL_calf_joint": -1.5,
        #     "RR_calf_joint": -1.5,
        # },
        # "dof_names": [
        #     "FR_hip_joint",
        #     "FR_thigh_joint",
        #     "FR_calf_joint",
        #     "FL_hip_joint",
        #     "FL_thigh_joint",
        #     "FL_calf_joint",
        #     "RR_hip_joint",
        #     "RR_thigh_joint",
        #     "RR_calf_joint",
        #     "RL_hip_joint",
        #     "RL_thigh_joint",
        #     "RL_calf_joint",
        # ],
        "default_joint_angles": {  # [rad]
            "torso_to_abduct_fl_j": 0.0,
            "torso_to_abduct_fr_j": 0.0,
            "torso_to_abduct_hr_j": 0.0,
            "torso_to_abduct_hl_j": 0.0,
            "abduct_fl_to_thigh_fl_j": 0.8,
            "abduct_fr_to_thigh_fr_j": 0.8,
            "abduct_hr_to_thigh_hr_j": 1.0,
            "abduct_hl_to_thigh_hl_j": 1.0,
            "thigh_fl_to_knee_fl_j": -1.5,
            "thigh_fr_to_knee_fr_j": -1.5,
            "thigh_hr_to_knee_hr_j": -1.5,
            "thigh_hl_to_knee_hl_j": -1.5,
        },
        "dof_names": [
            "torso_to_abduct_fr_j",
            "abduct_fr_to_thigh_fr_j",
            "thigh_fr_to_knee_fr_j",
            "torso_to_abduct_fl_j",
            "abduct_fl_to_thigh_fl_j",
            "thigh_fl_to_knee_fl_j",
            "torso_to_abduct_hr_j",
            "abduct_hr_to_thigh_hr_j",
            "thigh_hr_to_knee_hr_j",
            "torso_to_abduct_hl_j",
            "abduct_hl_to_thigh_hl_j",
            "thigh_hl_to_knee_hl_j",
        ],
        # PD
        "kp": 20.0,
        "kd": 0.5,
        # termination
        "termination_if_roll_greater_than": 10,  # degree
        "termination_if_pitch_greater_than": 10,
        # base pose
        "base_init_pos": [0.0, 0.0, 0.42],
        "base_init_quat": [1.0, 0.0, 0.0, 0.0],
        "episode_length_s": 20.0,
        "resampling_time_s": 4.0,
        "action_scale": 0.25,
        "simulate_action_latency": True,
        "clip_actions": 100.0,
    }
    obs_cfg = {
        "num_obs": 45,
        "obs_scales": {
            "lin_vel": 2.0,
            "ang_vel": 0.25,
            "dof_pos": 1.0,
            "dof_vel": 0.05,
        },
    }
    reward_cfg = {
        "tracking_sigma": 0.25,
        "base_height_target": 0.3,
        "feet_height_target": 0.075,
        "reward_scales": {
            "tracking_lin_vel": 1.0,
            "tracking_ang_vel": 0.2,
            "lin_vel_z": -1.0,
            "base_height": -50.0,
            "action_rate": -0.005,
            "similar_to_default": -0.1,
        },
    }
    command_cfg = {
        "num_commands": 3,
        "lin_vel_x_range": [0.5, 0.5],
        "lin_vel_y_range": [0, 0],
        "ang_vel_range": [0, 0],
    }

    return env_cfg, obs_cfg, reward_cfg, command_cfg


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("-e", "--exp_name", type=str, default="mini_cheetah-walking")
    parser.add_argument("-B", "--num_envs", type=int, default=4096)
    parser.add_argument("--max_iterations", type=int, default=100)
    args = parser.parse_args()

    gs.init(logging_level="warning")

    log_dir = f"logs/{args.exp_name}"
    env_cfg, obs_cfg, reward_cfg, command_cfg = get_cfgs()
    train_cfg = get_train_cfg(args.exp_name, args.max_iterations)

    if os.path.exists(log_dir):
        shutil.rmtree(log_dir)
    os.makedirs(log_dir, exist_ok=True)

    env = MiniCheetahEnv(
        num_envs=args.num_envs, env_cfg=env_cfg, obs_cfg=obs_cfg, reward_cfg=reward_cfg, command_cfg=command_cfg
    )

    runner = OnPolicyRunner(env, train_cfg, log_dir, device="cuda:0")

    pickle.dump(
        [env_cfg, obs_cfg, reward_cfg, command_cfg, train_cfg],
        open(f"{log_dir}/cfgs.pkl", "wb"),
    )

    runner.learn(num_learning_iterations=args.max_iterations, init_at_random_ep_len=True)


if __name__ == "__main__":
    main()

"""
# training
python examples/locomotion/mini_cheetah_train.py
"""

これで実行すると学習を行うことができます。

動作確認

go2の動作確認用スクリプトgo2_eval.pyをコピーし、mini_cheetah_eval.pyを作成します。
こちらは、上記で変更したクラス名を修正するのみです。
以下は動作確認用のmini_cheetah_eval.pyになります。

mini_cheetah_eval.py

import argparse
import os
import pickle

import torch
from mini_cheetah_env import MiniCheetahEnv
from rsl_rl.runners import OnPolicyRunner

import genesis as gs


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("-e", "--exp_name", type=str, default="mini_cheetah-walking")
    parser.add_argument("--ckpt", type=int, default=100)
    args = parser.parse_args()

    gs.init()

    log_dir = f"logs/{args.exp_name}"
    env_cfg, obs_cfg, reward_cfg, command_cfg, train_cfg = pickle.load(open(f"logs/{args.exp_name}/cfgs.pkl", "rb"))
    reward_cfg["reward_scales"] = {}

    env = MiniCheetahEnv(
        num_envs=1,
        env_cfg=env_cfg,
        obs_cfg=obs_cfg,
        reward_cfg=reward_cfg,
        command_cfg=command_cfg,
        show_viewer=True,
    )

    runner = OnPolicyRunner(env, train_cfg, log_dir, device="cuda:0")
    resume_path = os.path.join(log_dir, f"model_{args.ckpt}.pt")
    runner.load(resume_path)
    policy = runner.get_inference_policy(device="cuda:0")

    obs, _ = env.reset()
    with torch.no_grad():
        while True:
            actions = policy(obs)
            obs, _, rews, dones, infos = env.step(actions)


if __name__ == "__main__":
    main()

"""
# evaluation
python examples/locomotion/mini_cheetah_eval.py -e mini-cheetah-walking -v --ckpt 100
"""

以下は今回の結果になります。

Genesisで学習してmini_cheetah歩いた
（後ろ向き） pic.twitter.com/zOwnb2JPDX
— タマル (@tamarn5G) January 3, 2025

歩き方の修正

現在後ろ向きに歩いているのと、後ろ足が内股になっているのでそれらを修正します。
デフォルトの姿勢とのズレに対するペナルティーがあることから、初期関節角がおかしいため、後ろ向きの歩行に見えています。
（実際は前に進んでいるが、関節が逆向きになっている）
そのため、mini_cheetah_train.pyのデフォルトの関節角度を調整します。

mini_cheetah_train.py

"default_joint_angles": {  # [rad]
            "torso_to_abduct_fl_j": 0.0,
            "torso_to_abduct_fr_j": 0.0,
            "torso_to_abduct_hr_j": 0.0,
            "torso_to_abduct_hl_j": 0.0,
            "abduct_fl_to_thigh_fl_j": -0.8,
            "abduct_fr_to_thigh_fr_j": -0.8,
            "abduct_hr_to_thigh_hr_j": -0.8,
            "abduct_hl_to_thigh_hl_j": -0.8,
            "thigh_fl_to_knee_fl_j": 1.5,
            "thigh_fr_to_knee_fr_j": 1.5,
            "thigh_hr_to_knee_hr_j": 1.5,
            "thigh_hl_to_knee_hl_j": 1.5,
        },

結果は以下のようになりました。

デフォルトの関節角度調節したら前向きに歩いた pic.twitter.com/dqNUTkJYQc
— タマル (@tamarn5G) January 3, 2025

おわりに

Genesisでmini_cheetahの歩容生成ができたので、次は学習したPolicyを使って、Gazeboで動作させてみたいです。
また、これを機に強化学習をちゃんと学ぼうと思います。