AGILE#

A Generic Isaac-Lab based Engine for humanoid loco-manipulation learning.

AGILE is an external Isaac Lab project that showcases how to build a full RL training pipeline on top of Isaac Lab, from task design and training to evaluation and sim-to-real deployment.

Booster T1 – Stand-Up Booster T1 – Velocity Tracking

Sim

Real

Sim

Real
Unitree G1 – Velocity-Height Tracking Unitree G1 – Sit-Down / Stand-Up

Sim

Real

Sim

Real
Unitree G1 – Teleoperation Unitree G1 – Dancing

Sim

Real

Sim

Real

Key Features#

  • Multi-Robot Support: Validated on Booster T1 and Unitree G1 with sim-to-real transfer

  • Teacher-Student Distillation: Train with privileged observations, distill to deployable student policies

  • Self-Contained Tasks: Each task config is a single file; MDP term functions are shared across tasks via a common library (agile/rl_env/mdp/)

  • Evaluation Framework: Random rollouts, deterministic scenarios, motion metrics, HTML reports, W&B integration

  • Sim-to-MuJoCo Transfer: Generic framework for cross-simulator policy validation

  • Remote Training: OSMO workflow support for cluster-based training, evaluation, and sweeps

Quick Start#

# Train a velocity tracking policy
python scripts/train.py --task Velocity-T1-v0 --num_envs 2048 --headless

# Evaluate the trained policy
python scripts/eval.py --task Velocity-T1-v0 --num_envs 32 --checkpoint <path>

See Getting Started for full installation and setup instructions.

Supported Tasks#

Category

Task IDs

Description

Locomotion

Velocity-T1-v0, Velocity-G1-History-v0

Lower-body velocity tracking on rough terrain

Locomotion + Height

Velocity-Height-G1-v0, distillation variants

Lower-body velocity + height tracking with teacher-student distillation

Stand Up

StandUp-T1-v0

Full-body recovery from fallen poses (unified whole-body policy)

Pick & Place

G1-PickPlace-Tracking-v0

Upper-body trajectory tracking with frozen lower-body locomotion policy

Whole-Body Motion Tracking

Tracking-Flat-G1-v0

Full-body motion imitation from reference trajectories (e.g., dancing)

Debug

Debug-G1-v0, Debug-T1-v0

Interactive GUI environments for debugging

See Task System for detailed task documentation.