LAV2_base_att

rsl-rl config for LAV2 attitude control with rotation-manifold policy head.

Mirrors lav2/runner/skrl/cfg/LAV2_base_att.py but adapted to rsl-rl v5 native dict format. Uses :class:~lav2.runner.rsl_rl.models.AttitudeMLPModel as the actor model class and Gaussian distribution for rotation action spaces.

Usage (Isaac Lab Hydra entry point)::

"rsl_rl_cfg_entry_point": "lav2.runner.rsl_rl.cfg.LAV2_base_att:LAV2AttitudePPORunnerCfg"

Usage (direct script)::

from lav2.runner.rsl_rl.cfg.LAV2_base_att import get_runner_cfg
cfg = get_runner_cfg(experiment_name="my_run")
runner = OnPolicyRunner(env, cfg, log_dir, device=device)

Classes:

Name	Description
LAV2AttitudePPORunnerCfg	Hydra-compatible config class for the attitude-control task.

Functions:

Name	Description
get_runner_cfg	Return a rsl-rl v5 native config dict for the attitude-control task.

LAV2AttitudePPORunnerCfg

Bases: RslRlOnPolicyRunnerCfg

Hydra-compatible config class for the attitude-control task.

get_runner_cfg

get_runner_cfg(experiment_name: str = _EXPERIMENT_NAME, max_iterations: int = _MAX_ITERATIONS, projection_type: str | None = None, init_std: float = 0.135) -> dict

Return a rsl-rl v5 native config dict for the attitude-control task.

Parameters:

Name	Type	Description	Default
experiment_name	str	Logging directory name.	_EXPERIMENT_NAME
max_iterations	int	Number of PPO iterations.	_MAX_ITERATIONS
projection_type	str | None	Explicit projection class ("euler", "rotvec", "quat", "quat_plus", "quat_offset", "quat_plus_offset", "rotmat", "rotmat_offset"). When None, auto-detected from the environment action dimension.	None
init_std	float	Initial standard deviation for the Gaussian policy. Default 0.135 (matches so3_primer baseline log_std=-2.0).	0.135