Reinforcement Learning

Courses & Resources

Spinning Up in Deep RL

OpenAI 强化学习入门教程

OpenAI

intermediateSelf-paced

Isaac Gym: GPU RL Training

使用 Isaac Gym 进行大规模并行 RL 训练

NVIDIA

advanced4 hours

Code Demos

PPO with Isaac Gym

GPU Required

Run in Playground

使用 Isaac Gym 训练四足机器人行走策略

python

import isaacgym
from isaacgym import gymapi, gymtorch
import torch

# Initialize Isaac Gym
gym = gymapi.acquire_gym()
sim_params = gymapi.SimParams()
sim_params.use_gpu_pipeline = True

# Create simulation
sim = gym.create_sim(0, 0, gymapi.SIM_PHYSX, sim_params)

# Load robot asset
asset_root = "./assets"
asset_file = "anymal_c/urdf/anymal_c.urdf"
asset = gym.load_asset(sim, asset_root, asset_file)

print("Isaac Gym environment ready for RL training!")
print(f"Number of bodies: {gym.get_asset_rigid_body_count(asset)}")
print(f"Number of DOFs: {gym.get_asset_dof_count(asset)}")