Update skrl's agent configurations in the Isaac Lab template

[Toni-SM]

Description

Update skrl's agent configurations in the Isaac Lab template

Type of change

• Bug fix (non-breaking change which fixes an issue)

Checklist

• I have read and understood the contribution guidelines
• I have run the pre-commit checks with ./isaaclab.sh --format
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• I have updated the changelog and the corresponding version in the extension's config/extension.toml file
• I have added my name to the CONTRIBUTORS.md or my name already exists there

[greptile-apps]

Greptile Summary

This PR updates the Isaac Lab template skrl agent configuration files (AMP, IPPO, MAPPO) to align with the current skrl API, renaming and restructuring preprocessor and reward-scale fields.

• skrl_amp_cfg: Renames state_preprocessor→observation_preprocessor, amp_state_preprocessor→amp_observation_preprocessor, task/style_reward_weight→task/style_reward_scale (with style_reward_scale bumped from 1.0 to 2.0), and removes the now-redundant discriminator_reward_scale field.
• skrl_ippo_cfg: Swaps state_preprocessor: RunningStandardScaler for observation_preprocessor: RunningStandardScaler + an explicit state_preprocessor: null, matching the skrl IPPO API.
• skrl_mappo_cfg: Corrects the value-model input from OBSERVATIONS to STATES (required for the centralized critic in MAPPO), adds observation_preprocessor, and removes the obsolete shared_state_preprocessor field — all consistent with existing task-level MAPPO configs in the repo.

Confidence Score: 5/5

Safe to merge — all three template configs are updated to match the new skrl API and are already consistent with the actual task-level YAML configs in the repo.

Every changed field in all three templates was cross-validated against both the skrl documentation and the existing production task configs (skrl_walk_amp_cfg.yaml, cart_double_pendulum skrl_mappo/ippo_cfg.yaml, shadow_hand_over skrl_mappo_cfg.yaml). The value model input change in MAPPO from OBSERVATIONS to STATES is semantically correct for the centralized-critic design.

No files require special attention.

Important Files Changed

Filename	Overview
tools/template/templates/agents/skrl_amp_cfg	Renames state_preprocessor→observation_preprocessor, amp_state_preprocessor→amp_observation_preprocessor, task/style reward weight→scale fields, removes discriminator_reward_scale, and adjusts style_reward_scale from 1.0 to 2.0 — all consistent with the updated skrl AMP API and matching the existing humanoid_amp task configs.
tools/template/templates/agents/skrl_ippo_cfg	Replaces state_preprocessor with observation_preprocessor (null state_preprocessor) to match the updated skrl IPPO API; consistent with cart_double_pendulum's skrl_ippo_cfg.yaml.
tools/template/templates/agents/skrl_mappo_cfg	Switches value model input from OBSERVATIONS to STATES (correct for centralized MAPPO critics), adds observation_preprocessor, removes shared_state_preprocessor — matches both cart_double_pendulum and shadow_hand_over MAPPO configs.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    subgraph AMP["skrl_amp_cfg (AMP agent)"]
        A1[observation_preprocessor: RunningStandardScaler]
        A2[state_preprocessor: null]
        A3[amp_observation_preprocessor: RunningStandardScaler]
        A4[task_reward_scale: 0.0]
        A5[style_reward_scale: 2.0]
    end

    subgraph IPPO["skrl_ippo_cfg (IPPO agent)"]
        B1[observation_preprocessor: RunningStandardScaler]
        B2[state_preprocessor: null]
        B3[value_preprocessor: RunningStandardScaler]
    end

    subgraph MAPPO["skrl_mappo_cfg (MAPPO agent)"]
        C1[Policy model: input=OBSERVATIONS]
        C2[Value model: input=STATES]
        C3[observation_preprocessor: RunningStandardScaler]
        C4[state_preprocessor: RunningStandardScaler]
    end

    AMP -->|"Renamed from state_preprocessor"| A1
    AMP -->|"Renamed from amp_state_preprocessor"| A3
    AMP -->|"Renamed from task/style_reward_weight"| A4
    IPPO -->|"Renamed from state_preprocessor"| B1
    MAPPO -->|"Changed from OBSERVATIONS"| C2
    MAPPO -->|"Added; removed shared_state_preprocessor"| C3

Loading

_{Reviews (1): Last reviewed commit: "Update skrl's agents configurations in t..." | Re-trigger Greptile}